تقييم دقة بيانات التبخر والنتح المستندة إلى الأقمار الصناعية من OpenET لدعم تطبيقات إدارة الموارد المائية والأراضي Assessing the accuracy of OpenET satellite-based evapotranspiration data to support water resource and land management applications

المجلة: Nature Water، المجلد: 2، العدد: 2
DOI: https://doi.org/10.1038/s44221-023-00181-7
تاريخ النشر: 2024-01-15

تقييم دقة بيانات التبخر والنتح المستندة إلى الأقمار الصناعية من OpenET لدعم تطبيقات إدارة الموارد المائية والأراضي

تاريخ الاستلام: 21 يونيو 2023
تم القبول: 30 نوفمبر 2023
نُشر على الإنترنت: 15 يناير 2024

تحقق من التحديثات

الملخص

جون م. فولك © جاستن ل. هنتنغتون فورست س. ميلتون ريتشارد ألين مارثا أندرسون جوشوا ب. فيشر © أيşe كيلتش أندرسون روهوف © جابرييل ب. سناي بليك ماينور تشارلز مورتون توماس أوت لي جونسون (1) برونو كوميني دي أندرادي ويل كارارا كونور ت. دوهيرتي كريستيان دانكرلي ماكنزي فريدريش ألبرتو غوزمان كريستوفر هاين غريغوري هالفيرسون يانغ هوي كانغ (ب كايل كنيبر (ب ليوناردو لايبيلت صموئيل أورتيغا-سالازار كريستوفر بيرسون جابرييل إي. إل. بارريش آدم بوردي بيتر ريفيل تيانشين وانغ ويون يانغ

الملخص

تقدم بيانات التبخر والنتح المستشعرة عن بُعد (ET) إمكانيات قوية لدعم الأساليب المعتمدة على البيانات لإدارة المياه المستدامة. ومع ذلك، يحتاج الممارسون إلى تقييمات دقيقة وصارمة لمدى دقة هذه البيانات. تم تطوير نظام OpenET، الذي يتضمن مجموعة من ستة نماذج استشعار عن بُعد، لزيادة الوصول إلى بيانات ET على مستوى الحقل (30 م) للولايات المتحدة المتجاورة. هنا نقارن مخرجات OpenET ضد بيانات من 152 محطة في الموقع، تتكون أساسًا من أبراج تدفق التباين الدوامي، المنتشرة عبر الولايات المتحدة المتجاورة. متوسط الخطأ المطلق في مواقع الأراضي الزراعية لقيمة مجموعة OpenET هو 15.8 مم في الشهر ( متوسط خطأ الانحياز هو -5.3 مم في الشهر ( ) و هو 0.9. تظهر النتائج لمناطق الشجيرات والمواقع الحرجية تباينًا أكبر بين النماذج ودقة أقل مقارنة بالأراضي الزراعية. تشير الدقة العالية وتوافق النماذج المتعددة عبر الأراضي الزراعية إلى فائدة نهج تجميع النماذج، وتعزز الثقة بين ممارسي بيانات التبخر، بما في ذلك مجتمع إدارة الموارد المائية الزراعية.

تعتبر بيانات التبخر والنتح (ET) الدقيقة ضرورية لتقييم توازن الطاقة والمياه السطحية، ودورة الكربون، وإدارة الموارد المائية. ET هو مجموع تدفق بخار الماء من التربة (التبخر) ومن خلال النباتات (النتح) إلى الغلاف الجوي. يشكل ET ثاني أكبر مكون من موازنة المياه الأرضية، بعد الهطول. لقد زادت فائدة رسم خرائط ET المتجاورة مكانيًا، لا سيما فوق الأراضي الزراعية المروية، بسبب الجفاف وتغير المناخ وارتفاع معدلات سحب المياه البشرية واستهلاكها الزراعي، مما يترك
العديد من المياه الجوفية وخزانات المياه في غرب الولايات المتحدة عند أدنى مستوياتها على الإطلاق تقدم تقنية الاستشعار عن بُعد المعتمدة على الأقمار الصناعية لتقدير التبخر (RSET) نهجًا قويًا لرسم خرائط التبخر عبر مناطق جغرافية واسعة على مقاييس زمنية شبه مستمرة. حتى وقت قريب، كانت توفر بيانات RSET على مقاييس مكانية ذات صلة بإدارة موارد المياه محدودًا بسبب التكاليف ومتطلبات الحوسبة.
OpenET تستخدم ستة نماذج RSET متطورة تعتمد على الأقمار الصناعية، وهي ALEXI/DisALEXI ، eeMETRIC جي سيبال ، PT-JPL سيمز و SSEBop التي تم تطبيقها وتقييمها على نطاق واسع في الولايات
الشكل 1 | خريطة مواقع قياس ET في الموقع. خريطة لمواقع محطات ET في الموقع المستخدمة لتقييم OpenET، بما في ذلك نوع الغطاء الأرضي العام ومناطق المناخ وفقًا لنظام كوبن-جيجر (KG). تمثل المناطق البيضاء مناطق المناخ التي لم تحتوي على أي مواقع للزراعة وتم استبعادها من التحليل.
تُعرّف اختصارات مناطق المناخ على النحو التالي: السهوب شبه الجافة الباردة والحارة (Bsk + Bsh)؛ الصحراء الحارة والباردة (Bwh + Bwk)؛ المناخ الرطب شبه الاستوائي (Cfa)؛ البحر الأبيض المتوسط ذو الصيف الحار والدافئ (Csa + Csb)؛ والمناخ القاري الرطب ذو الصيف الحار والدافئ (Dfa + Dfb).
دول لمجموعة من تطبيقات إدارة المياه والزراعة. يتم تطبيق النماذج على منصة جوجل إيرث إنجن السحابية. لتوفير بيانات ET التاريخية واللحظية تقريبًا على مقاييس دون الحقل بكسل) فوق غرب الولايات المتحدة خمسة من نماذج RSET تقيد مكونات توازن الطاقة السطحية (SEB) باستخدام درجة حرارة سطح الأرض (LST) المستمدة أساسًا من مجموعة لاندسات 2، جنبًا إلى جنب مع بيانات الطقس الموزعة جغرافيًا، وبيانات تغطية الأرض. النموذج السادس، SIMS، يفترض ظروف ري جيدة ويحسب معاملات المحاصيل بناءً على كثافة النبات، المستمدة من قيم انعكاس السطح من الأقمار الصناعية، بالإضافة إلى نموذج توازن المياه في التربة الموزع جغرافيًا. تم استخدام النماذج المكونة لـ OpenET من قبل مديري المياه، والمزارعين، والمنظمات الحكومية لجدولة الري، ومحاسبة المياه وتخصيصها، وإدارة حقوق المياه. . توفر منصة OpenET مستوى غير مسبوق من الوصول إلى بيانات RSET من خلال واجهة استكشاف البيانات العامة عبر الإنترنت – بما في ذلك استعلام ET الساتلي ضمن حدود الحقول المرسومة بشكل فردي. تعمل جميع نماذج RSET الستة في OpenET تلقائيًا، بما في ذلك أي معايرات مطلوبة، مما يسمح بإجراء حسابات سريعة لأكثر من 100,000 صورة من لاندسات تمت معالجتها حتى الآن عبر 23 ولاية من الولايات الغربية في الولايات المتحدة المتجاورة. مع تزايد عدد تطبيقات بيانات RSET لإدارة الموارد الأرضية والمائية المستدامة، من المهم أن يكون لدى الممارسين معلومات حول دقة بيانات RSET عبر أنواع تغطية الأرض، والمناطق المناخية، وممارسات الإنتاج الزراعي. .
في هذه التحليل، نقدم تقييمًا مرجعيًا واسع النطاق لدقة بيانات OpenET باستخدام مجموعة بيانات مؤرشفة عامة تم تنسيقها جيدًا من قياسات ET في الموقع من 152 محطة (141 نظام تباين دوامي (EC)، 7 أنظمة نسبة بوان، و4 ليزيمترات)، عبر مجموعة متنوعة من المناطق والمناخات وأنواع تغطية الأرض. ، التي تتكون بشكل جماعي من سنوات من بيانات نموذج القياس ET المزدوجة (الشكل 1). تُعتبر تقنية EC عمومًا أفضل طريقة متاحة للقياس المستمر لتدفق الطاقة والحرارة في الموقع على مقاييس مكانية تقترب من الاسترجاعات المعتمدة على الأقمار الصناعية. على الرغم من أننا نعترف بعدم اليقين المرتبط بالبيانات وقدمنا جهودًا لتقليلها . بالإضافة إلى تقييم دقة النماذج الفردية، قمنا بتقييم قيمة ET لمجموعة OpenET، المحسوبة كمتوسط لجميع النماذج بعد تحديد وإزالة ما يصل إلى حالتين شاذتين باستخدام الوسيط المطلق.
نهج الانحراف (MAD) توليد قيمة جماعية هو تقنية مستخدمة على نطاق واسع لدمج المخرجات من نماذج متنوعة، كل منها له سلوكه الخاص. وخطأ عشوائي . كما أنه يسهل التطبيقات مثل جدولة الري وإدارة حقوق المياه، حيث يحتاج الممارسون إلى قيمة واحدة لاستخدامها في إدارة موارد المياه تسمح مجموعة بيانات التدفق في الموقع المؤرشفة علنًا بإعادة الإنتاج ومعايرة إصدارات نموذج OpenET المستقبلية أو بيانات RSET الأخرى.
تم الحصول على بيانات ET المحسوبة من قياسات الميكرو ميتورولوجيا في مواقع EC من مجموعة متنوعة من المصادر، principalmente AmeriFlux. الجدول التكميلي 1 يوفر قائمة كاملة بالمحطات المستخدمة في الدراسة بما في ذلك نوع تغطية الأرض، المحققين الرئيسيين في الموقع، معرفات الكائنات الرقمية (DOIs) وبيانات وصفية أخرى. تم معالجة بيانات التدفق بعناية بعد جمعها، بما في ذلك ملء الفجوات، وفحص أخطاء إغلاق توازن الطاقة واكتمال البيانات، وتقييمات جودة البيانات البصرية. تم تضمين بيانات التدفق التي اجتازت مراقبة الجودة وأظهرت أخطاء محدودة في إغلاق توازن الطاقة في الدراسة وخضعت لتصحيح الإغلاق وفقًا لنهج FLUXNET2015/ONEFlux للتدفقات المتوسطة اليومية. نحن نشير إلى بيانات EC باسم ‘ECET’ في جميع أنحاء المقال. تم اعتبار بيانات ECET المغلقة الأكثر تمثيلاً للواقع. لتجميع بكسلات RSET للمقارنة مع ECET، تم تطوير بصمات التدفق لكل محطة. بصمات التدفق هي خرائط ثنائية الأبعاد لمدى منطقة مصدر المحطة، أي المنطقة على الأرض التي تساهم في التدفقات التي تقيسها أدوات البرج. يرجى الرجوع إلى الطرق و Volk وآخرون. للحصول على تفاصيل حول معالجة بيانات التدفق وطرق رسم البصمة المستخدمة. يتم تقديم مناقشة إضافية حول عدم اليقين في بيانات التبادل الكلي والخطوات المتخذة للحد من هذا عدم اليقين في المناقشة التكميلية 1. يتم تقديم نظرة عامة على نماذج ET المدفوعة بالأقمار الصناعية في مجموعة OpenET في الطرق.
تركز المناقشة حول النتائج الإحصائية التي تلي ذلك على المقارنات بين إجمالي ECET و RSET الشهري. على الرغم من أنه تم إجراء تقييمات دقة باستخدام بيانات يومية (تاريخ المرور) وإجمالي ET الشهري المجمّع لفترات موسم النمو والفترات السنوية، تركز مناقشتنا على النتائج الشهرية لعدة أسباب: إن ET الشهري له فائدة في المحاسبة المائية والتخطيط على المدى الطويل؛ كما أن الشكوك في بيانات EC بسبب الإغلاق وعوامل أخرى تقلل.
الجدول 1 | إحصائيات ملخصة بين ET الشهري المودل والملاحظ لمواقع الأراضي الزراعية
نوع تغطية الأرض إحصائية مجموعة ديزاكسي eeMETRIC جي سيبال PT-JPL سيمز SSEBop مواقع نقاط البيانات
جميع المحاصيل، متوسط ET للمحطة 91 (مم في الشهر) منحدر 0.92 0.92 0.95 0.85 0.91 0.99 0.95 53 1,652
MBE (مم) -5.27 (-5.8%) -7.72 (-8.4%) -2.44 (-2.7%) -12.18 (-13.3%) -2.9 (-3.2%) 4.32 (4.7%) -6.08 (-6.7%) ٤٤ 1,638
MAE (مم) 15.84 (17.3%) 19.91 (21.8%) 21.23 (23.2%) 22.69 (24.8%) 18.12 (19.8%) 17.93 (19.6%) 22.4 (24.5%) ٤٤ 1,638
جذر متوسط مربع الخطأ (مم) 20.44 (22.4%) 25.35 (27.7%) ٢٦.٩٧ (٢٩.٥٪) ٢٩.٠٥ (٣١.٨٪) 23.67 (25.9%) 23.1 (25.3%) 27.72 (30.3%) ٤٤ 1,638
0.9 0.86 0.83 0.83 0.87 0.86 0.85 53 1,652
المحاصيل السنوية، متوسط ET للمحطة 85 (مم في الشهر) منحدر 0.93 0.92 0.98 0.85 0.9 1.01 0.92 42 ١٤٤٦
MBE (مم) -5.11 (-6.0%) -8.18 (-9.6%) 0.23 (0.3%) -12.38 (-14.6%) -3.77 (-4.4%) 6.27 (7.4%) -9.13 (-10.7%) ٣٦ ١٤٣٦
MAE (مم) 15.26 (17.9%) ٢٠.٠٩ (٢٣.٦٪) 20.44 (24.0%) 22.52 (26.5%) 17.0 (20.0%) 17.48 (20.5%) 21.93 (25.8%) ٣٦ ١٤٣٦
جذر متوسط مربع الخطأ (مم) 19.71 (23.2%) 25.68 (30.2%) ٢٦.١٧ (٣٠.٨٪) 28.67 (33.7%) 22.31 (26.2%) 22.49 (26.4%) 27.14 (31.9%) ٣٦ ١٤٣٦
0.9 0.84 0.83 0.82 0.87 0.85 0.84 42 ١٤٤٦
البساتين، متوسط محطة ET هو 126 (مم في الشهر) منحدر 0.87 0.88 0.81 0.84 0.88 0.93 0.97 ٥ 141
MBE (مم) -11.9 (-9.4%) -11.02 (-8.7%) -20.66 (-16.4%) -15.11 (-12.0%) -7.39 (-5.8%) -3.47 (-2.7%) -3.69 (-2.9%) ٥ 141
MAE (مم) 21.18 (16.8%) 22.2 (17.6%) 28.19 (22.3%) 24.9 (19.7%) 24.43 (19.3%) 22.95 (18.2%) 20.18 (16.0%) ٥ 141
جذر متوسط مربع الخطأ (مم) 27.89 (22.1%) 27.86 (22.1%) 35.26 (27.9%) 32.77 (25.9%) 31.67 (25.1%) 30.49 (24.1%) 27.26 (21.6%) ٥ 141
0.91 0.89 0.89 0.88 0.88 0.86 0.89 ٥ 141
كروم العنب، متوسط المحطة ET هو 112 (مم في الشهر) منحدر 1.02 1.02 0.95 0.95 1.09 0.92 1.25 ٣ 61
MBE (مم) 5.27 (4.7%) 4.99 (4.5%) -4.62 (-4.1%) -3.76 (-3.4%) 17.95 (16.0%) -7.71 (-6.9%) 31.88 (28.5%) ٣ 61
MAE (مم) 13.66 (12.2%) 13.01 (11.6%) 18.73 (16.7%) 20.72 (18.5%) 21.53 (19.2%) 14.43 (12.9%) 33.22 (29.7%) ٣ 61
جذر متوسط مربع الخطأ (مم) 16.23 (14.5%) 15.87 (14.2%) 22.1 (19.7%) 27.2 (24.3%) 27.19 (24.3%) 17.34 (15.5%) ٣٦.٥٦ (٣٢.٦٪) ٣ 61
0.9 0.9 0.81 0.73 0.83 0.88 0.84 ٣ 61

من متوسط التدفق المغلق للبرج المثقل ET. لاحظ أنه كان هناك ثلاثة مواقع إضافية لمحاصيل الخضروات مدرجة في مجموعة المحاصيل المجمعة، والتي بمفردها لم تستوفِ متطلبات بياناتنا للتحليلات الإحصائية. .
على المقياس الزمني الشهري (مقارنةً باليومي)، يوفر OpenET مباشرةً ET اليومية والشهرية، بالإضافة إلى خدمات البيانات التي تتيح للمستخدمين حساب ET في فترات تجميع أخرى. يتم تقديم نتائج الدقة لفترات زمنية يومية وشهرية وموسمية وسنوية في الجداول التكميلية 2-6، وينبغي استشارة مقاييس الدقة لفترات زمنية يومية لتطبيقات بيانات ET في فترات زمنية تتراوح بين 1-15 يومًا. تم استخدام خمسة مقاييس إحصائية معروفة لتقييم دقة OpenET (للمعادلات، انظر الطرق): ميل الانحدار الخطي المفروض من خلال الأصل الذي يقيس التحيز (الميل)، خطأ التحيز المتوسط (MBE)، خطأ القيمة المطلقة المتوسطة (MAE)، خطأ الجذر التربيعي المتوسط (RMSE) ومعامل التحديد. نتائج الانحدار مع تقاطع غير صفري للبيانات الشهرية موضحة في الجدول التكميلية 7.

الأداء عبر جميع مواقع تدفق الزراعة

من بين جميع أنواع تغطية الأراضي العامة التي تم أخذ عينات منها، أظهرت نماذج OpenET أقوى توافق مع ECET المجمعة في البيئات الزراعية. بالنسبة لـ 44 موقعًا زراعيًا مجتمعة، أظهرت eeMETRIC وSIMS وPT-JPL أقل انحياز من حيث MBE، جميعها أقل من -4.5 مم في الشهر أو من متوسط ECET (الجدول 1). كانت قيمة المجموعة تحتوي على انحياز طفيف أعلى بمقدار -5.3 مم في الشهر، أو من متوسط ECET. كانت قيمة المجموعة أفضل من كل نموذج فردي من حيث MAE 15.9 مم في الشهر ( من متوسط ECET)، RMSE 20.4 مم في الشهر ( ) و ( 0.90 ). بالمقارنة، تراوحت MAE من النماذج الفردية بين 17.9 إلى 22.7 مم، و RMSE بين 23.1 إلى 29.1 مم شهريًا و من 0.83 إلى 0.87 مع أصغر الأخطاء من PT-JPL وSIMS وDisALEXI.
كانت بيانات ET من نماذج RSET الفردية مرتبطة عمومًا بشكل خطي بـ ECET، حيث أظهرت PT-JPL وSIMS بعض الانحناءات بسبب التحيزات المتغيرة موسميًا (الشكل 2). وقد قدرت العديد من النماذج ET خلال الموسم البارد بشكل أقل مقارنة بـ ECET، مما أدى إلى التحيز المنخفض قليلاً في قيمة ET الجماعية (الجدول 2). للتحقيق في التباين الموسمي في دقة النموذج، قمنا بتجميع جميع الأزواج الشهرية.
ET المقاسة بواسطة النموذج لتوليد المناخيات الشهرية لتصنيفات الغطاء الأرضي الرئيسية (الشكل 3 والأشكال البيانية الموسعة 1-5). النطاق بين ECET غير المغلق والمغلق يوفر مقياسًا واحدًا لعدم اليقين في البيانات الميدانية. .
في معظم الأشهر، كانت قيمة ET في مجموعة النماذج المتعددة محصورة بشكل جيد بين المتوسط المغلق وغير المغلق ECET لمواقع الأراضي الزراعية، بينما أظهرت الأعضاء الفردية في المجموعة انحرافًا موسميًا أكبر. في الربيع، كانت SSEBop وeeMETRIC تقدر ET غير المغلق بشكل أقل، بينما كانت SIMS تقدر ET المغلق بشكل زائد، ربما بسبب الافتراض بوجود ظروف مروية جيدًا. في أشهر الصيف ذروتها، كانت معظم النماذج متوافقة بشكل جيد مع ECET المغلق، مع انحراف منخفض لـ geeSEBAL وPT-JPL. في سبتمبر وأكتوبر، عندما تنخفض معدلات ET الفعلية بسرعة، كانت عدة نماذج منحرفة بشكل مرتفع، باستثناء DisALEXI وgeeSEBAL، اللذان تتبعا القيم غير المغلقة بشكل أقرب. إن التوافق الأعلى لـ RSET مع ECET خلال فترة ذروة الصيف يعد مشجعًا، حيث إنها فترة الري المكثف والاستخدام الاستهلاكي للمياه من خلال ET. أظهر اختبار لاحق أن DisALEXI وgeeSEBAL وSSEBop كانت لديها قيم ET الشهرية المتوسطة التي كانت مختلفة إحصائيًا (كأقل تقدير) عن المتوسط المغلق لـ ECET. لم تكن قيمة ET المتوسطة المجمعة لفترة النمو لجميع النماذج مختلفة عن المتوسط المغلق لـ ECET (الجداول التكميلية 8 و9).
تمت ترقية المناخيات الشهرية المستمدة من مواقع التدفق باستخدام بيانات من جميع بكسلات الأراضي الزراعية عبر نطاق OpenET الكامل (الشكل البياني الإضافي 6). وجدنا أنماط موسمية مماثلة وانحيازات نسبية للنموذج مشابهة لتلك التي تم تحديدها في مواقع التدفق، مما يمنح الثقة في تمثيل مقارنات ECET.

أثر فترة العينة على أداء النموذج

غالبًا ما تتحسن دقة النموذج مع زيادة فترة التجميع الزمني بسبب إلغاء الأخطاء. في الأراضي الزراعية، تحسنت مقاييس الدقة لمجموعة OpenET مع زيادة فترة التجميع.
الشكل 2 | النمذجة مقابل الملاحظة لمعدل التبخر الشهري في مواقع الأراضي الزراعية. مقارنة شهرية لجميع أزواج OpenET أعضاء المجموعة ET مقابل برج التدفق المغلق من جميع محطات الأراضي الزراعية لجميع أشهر السجل. تم تضمين نتيجة نموذج الانحدار الخطي الأقل تربيعًا لكل نموذج، مع فرضه عبر الأصل و .
من التواريخ اليومية (تاريخ العبور) إلى الشهرية إلى موسم النمو إلى الفترات السنوية (الجداول التكميلية 2-6). أظهرت نتائج المجموعة اليومية لمواقع الأراضي الزراعية المجمعة متوسط خطأ مطلق قدره ، و RMSE لـ متوسط ECET. في هذا الإطار الزمني، هناك زيادة في عدم اليقين في بيانات ECET بسبب التغيرات في الظروف الميكرو-مناخية وإغلاق توازن الطاقة، وكذلك في ET المستشعر عن بُعد بسبب احتمال تلوث السحب والأخطاء في تمثيل البصمة. يتم تقليل هذه الشكوك الجماعية عند التكامل إلى شهري (MAE من و RMSE لـ من ECET)، موسم النمو (MAE من و
جذر متوسط مربع الخطأ (RMSE) بنسبة 15.5% من ECET) ومياه السنة (MAE بنسبة 11.3% و RMSE بنسبة 12.3% من ECET) على فترات زمنية. لحسن الحظ، خلال فترات موسم النمو وجدنا خطأ أقل في إغلاق توازن الطاقة في بيانات EC. وهناك تغطية سحابية أقل في بيانات الأقمار الصناعية في غرب الولايات المتحدة مقارنةً بفترة عدم النمو. خلال الصيف، كان متوسط الخطأ المطلق الطبيعي اليومي (NMAE) في تواريخ العبور عادةً بين 5% و (الشكل التوضيحي 1)، ونسبة 7% و20% شهريًا (الشكل 4). نتوقع أن تكون فترات التجميع المخصصة بين 2 و15 يومًا لها دقة مشابهة أو محسنة قليلاً مقارنة بالنتائج اليومية التي تتغير موسميًا؛ قد تكون نتائج RSET من أسبوع إلى أسبوعين الأكثر فائدة لجدولة الري. .

الأداء بين المحاصيل السنوية والدائمة

تشكل المحاصيل السنوية، بما في ذلك القمح والذرة وفول الصويا والأرز وغيرها، الغالبية (80%) من مواقع الأراضي الزراعية في مجموعة بيانات OpenET ECET (الجدول التكميلي 1). مقارنة بالمحاصيل المعمرة، تميل المحاصيل السنوية إلى أن تكون لها مظلات أقصر وتغطية أكثر تجانسًا في مرحلة الذروة للنمو. مواقع المحاصيل السنوية في مجموعة بيانات تدفق OpenET هي في الغالب مروية، وتوزع عبر مجموعة من المناطق المناخية، مع كثافة أعلى في مناطق مثل البحر الأبيض المتوسط ووادي كاليفورنيا شبه الجاف، والمناطق القارية الرطبة في السهول العليا وسهول المسيسيبي الطينية (الشكل 1).
بالنسبة للمحاصيل السنوية، أظهرت كل من نماذج RSET في مجموعة OpenET انحيازًا صغيرًا ومستويات عالية من الدقة والضبط (الجدول 1). مشابهًا لجميع أنواع المحاصيل مجتمعة، تفوقت القيمة الجماعية للمحاصيل السنوية على النماذج الفردية من حيث MAE (15.3 مم في الشهر أو 17.9% من متوسط ECET) و RMSE (19.7 مم في الشهر أو 23.2% من متوسط ECET) و (0.9). من بين نماذج RSET، أظهرت eeMETRIC و PT-JPL أقل مقدار من MBE، حيث حقق PT-JPL و SIMS أعلى دقة من حيث MAE و RMSE.
عند تقسيم المحاصيل السنوية إلى فئات C3 وC4، نجد أن أنماط الموسم وأحجام متوسط الخطأ المطلق الجماعي (MAE) متشابهة على مدار العام (الشكل 4). يعكس متوسط الخطأ المطلق الطبيعي (NMAE) بشكل عام عكس منحنى استخدام المياه المميز لكل فئة، حيث تظهر محاصيل C3 منحنى موسمي أوسع من C4 وبالتالي NMAE أقل في بداية ونهاية الموسم. بينما القيم الأعلى لـ NMAE التي لوحظت خارج موسم النمو لجميع أنواع المحاصيل (الشكل 4) تشير بشكل أكبر إلى معدلات تبخر منخفضة بدلاً من خصائص خطأ النمذجة ذات الدلالة، قد تكون أخطاء موسم البرودة مبالغ فيها بشكل عام بسبب زيادة تغطية السحب، مما يزيد من الفترة الزمنية بين استرجاعات الأقمار الصناعية الخالية من السحب. يجب أن يساعد تحسين تكرار تصوير الأقمار الصناعية، بالإضافة إلى تقنيات دمج الوقت لتبخر المياه وملء الفجوات، في زيادة دقة OpenET خلال موسم عدم النمو (النقاش).
فئة أخرى من الاهتمام هي النباتات المعمرة الخشبية، التي تعتبر محاصيل ذات قيمة عالية وتطرح تحديات نمذجة مميزة. كانت بيانات تدفق الإدي عالية الجودة متاحة لثلاثة مزارع عنب، وثلاثة بساتين أشجار مكسرات، وبستان فواكه واحد، جميعها تقع في كاليفورنيا. تتميز مزارع الكروم والبساتين بوجود مظلات أطول وأكثر هيكلة، وغالبًا ما تحتوي على محاصيل تغطية بين الصفوف، وغالبًا ما يتم ري مزارع الكروم بشكل ناقص. تؤدي هذه الخصائص إلى تأثيرات الظل والبكسل المختلط في الاستشعار عن بعد في المستوى، والحاجة إلى الحساسية للتغيرات الصغيرة في إجهاد الكرمة لإبلاغ تطبيقات الري الناقص هي متطلبات نمذجة فريدة.
كان أداء نموذج RSET في مواقع الكروم الم sampled قويًا ومتسقًا عبر النماذج. تجاوزت دقة التجميع تلك الخاصة بالمحاصيل السنوية (الجدول 1 والشكل 4)، مع انحياز أقل (ميل 1.02 و MBE 5.3 مم في الشهر) وانحراف مطلق متوسط أقل و RMSE (13.7 و 16.2 مم في الشهر، على التوالي، أو 12.2% و 14.5% من متوسط ECET الشهري) و 0.90. أدت DisALEXI بشكل مشابه أو أفضل من المجموعة في مواقع تدفق الكروم، ربما بسبب نهجها ذو المصدرين في تقسيم تدفقات الحرارة بين الركيزة (بين الصفوف) والغطاء.
كانت الأداءات أكثر تنوعًا بين أعضاء الفرقة بالنسبة للبساتين مقارنة بأنواع المحاصيل الأخرى، وكانت التحيزات أكثر سلبية. قد يكون هذا مرتبطًا بتأثيرات الظل في النباتات الأطول والأكثر…
الجدول 2 | إحصائيات ملخصة بين ET الشهري المودل والملاحظ لمواقع الأراضي الزراعية المجمعة حسب منطقة المناخ
نوع تغطية الأرض إحصاء مجموعة ديزاكسي eeMETRIC جي سيبال PT-JPL سيمز SSEBop مواقع نقاط البيانات
بسك + بش (سافانا شبه جافة باردة وحارة)، متوسط تبخر المحطة 133 (مم في الشهر) منحدر 0.9 0.85 0.91 0.82 0.88 0.97 1 11 246
MBE (مم) -6.94 (-5.2%) -15.17 (-11.4%) -4.65 (-3.5%) -18.09 (-13.6%) -7.55 (-5.7%) 2.71 (2.0%) 2.89 (2.2%) 11 246
MAE (مم) 20.74 (15.6%) ٢٦.٩٣ (٢٠.٣٪) ٢٦.٩٤ (٢٠.٣٪) 30.31 (22.8%) 25.7 (19.3%) 22.99 (17.3%) ٢٤.٢٣ (١٨.٢٪) 11 246
جذر متوسط مربع الخطأ (مم) ٢٦.٣٨ (١٩.٨٪) ٣٤.٥٥ (٢٦.٠٪) 33.4 (25.1%) ٣٨.٠ (٢٨.٦٪) 32.49 (24.4%) 28.97 (21.8%) 30.99 (23.3%) 11 246
0.89 0.81 0.8 0.8 0.84 0.84 0.84 11 246
بوه + بوك (صحراء حارة وباردة)، متوسط محطة ET يبلغ 110 (مم في الشهر) منحدر 0.91 0.85 1.02 0.92 0.86 0.92 0.88 10 53
MBE (مم) -6.78 (-6.1%) -15.63 (-14.2%) 9.63 (8.7%) -8.19 (-7.4%) -7.77 (-7.0%) -3.2 (-2.9%) -12.34 (-11.2%) ٧ ٤٩
MAE (مم) 13.24 (12.0%) 21.21 (19.2%) 18.92 (17.1%) 19.13 (17.3%) 19.21 (17.4%) 13.62 (12.3%) 19.59 (17.8%) ٧ ٤٩
جذر متوسط مربع الخطأ (مم) 17.02 (15.4%) 25.78 (23.4%) 23.92 (21.7%) 23.51 (21.3%) 23.81 (21.6%) 16.07 (14.6%) 22.95 (20.8%) ٧ ٤٩
0.91 0.85 0.88 0.87 0.83 0.94 0.89 10 53
Cfa (رطب شبه استوائي)، متوسط ET للمحطة 75 (مم في الشهر) منحدر 1 1.03 1.03 0.99 0.93 1.15 0.91 11 232
MBE (مم) 2.15 (2.9%) 4.49 (6.0%) 3.65 (4.9%) 0.97 (1.3%) -0.88 (-1.2%) 14.98 (19.9%) -3.84 (-5.1%) ٨ 228
MAE (مم) 17.51 (23.3%) 20.17 (26.8%) 24.01 (31.9%) 20.06 (26.7%) 17.79 (23.6%) 22.71 (30.2%) 21.97 (29.2%) ٨ 228
جذر متوسط مربع الخطأ (مم) 23.76 (31.6%) ٢٦.١٨ (٣٤.٨٪) 31.88 (42.4%) 28.39 (37.7%) 23.62 (31.4%) 30.23 (40.2%) 28.76 (38.2%) ٨ 228
0.75 0.72 0.62 0.69 0.76 0.72 0.64 11 232
CSA + CSB (المتوسطات المناخية للبحر الأبيض المتوسط ذات الصيف الحار والدافئ)، متوسط ET للمحطة هو 94 (ملم في الشهر) منحدر 0.95 0.96 0.99 0.81 1.01 0.93 0.99 ٨ ٢٩٢
MBE (مم) -4.37 (-4.7%) -6.32 (-6.7%) -1.72 (-1.8%) -20.92 (-22.3%) 7.37 (7.9%) -1.34 (-1.4%) -2.9 (-3.1%) ٨ ٢٩٢
MAE (مم) 13.32 (14.2%) 18.65 (19.9%) 17.14 (18.3%) 25.9 (27.6%) 17.17 (18.3%) 14.04 (15.0%) 23.94 (25.6%) ٨ ٢٩٢
جذر متوسط مربع الخطأ (مم) 16.52 (17.6%) 22.75 (24.3%) 21.48 (22.9%) 31.31 (33.4%) 21.97 (23.5%) 18.27 (19.5%) 27.92 (29.8%) ٨ ٢٩٢
0.93 0.87 0.88 0.85 0.87 0.88 0.84 ٨ ٢٩٢
دي إف إيه + دي إف بي (قاري رطب صيف حار وصيف دافئ)، متوسط ET للمحطة 67 (مم في الشهر) منحدر 0.9 0.93 0.94 0.86 0.87 1.02 0.9 ١٣ 829
MBE (مم) -8.0 (-11.9%) -8.07 (-12.0%) -6.74 (-10.0%) -10.55 (-15.7%) -5.72 (-8.5%) 4.88 (7.3%) -13.58 (-20.2%) 10 ٨٢٣
MAE (مم) 13.8 (20.5%) 15.68 (23.3%) 18.93 (28.2%) 17.88 (26.6%) 13.67 (20.3%) 15.35 (22.8%) 21.09 (31.4%) 10 ٨٢٣
جذر متوسط مربع الخطأ (مم) 17.85 (26.6%) 20.36 (30.3%) ٢٤.١ (٣٥.٨٪) 23.36 (34.7%) 18.89 (28.1%) 19.94 (29.7%) 25.9 (38.5%) 10 ٨٢٣
0.91 0.9 0.83 0.86 0.89 0.86 0.86 ١٣ 829
إحصائيات ملخصية شهرية متوسطة للمقارنات بين OpenET أعضاء المجموعة ET وبرج التدفق المغلق ET الشهري للمواقع الزراعية المجمعة حسب مناخ كوبن-جيجر
متوسط الوزن لإنتاجية التبخر من برج التدفق المغلق.
الأغصان المتجمعة بقوة، خاصة بالنسبة للنماذج التي تعتمد بشكل كبير على مدخلات درجة حرارة سطح الأرض. كانت القيمة الجماعية تحمل انحيازًا سلبيًا بمتوسط ميل قدره في الشهر، MAE 21.2 مم في الشهر ( من ECET) و RMSE 27.9 مم في الشهر (22.1% من ECET)، و من 0.91. كان لدى SSEBop وSIMS أقل انحياز من حيث الميل وMBE، وكان لدى SSEBop وDisALEXI أقل خطأ من حيث MAE وRMSE (الجدول 1). بينما يكون MAE في البساتين مرتفعًا في منتصف الموسم، فإن القيم المعيارية مشابهة لتلك الخاصة بالمحاصيل السنوية (الشكل 4).

تباين أداء النموذج عبر المناطق المناخية

لتحقيق في التباينات في أداء OpenET عبر مناخات مختلفة، تم تجميع مقاييس دقة الأراضي الزراعية حسب مناطق المناخ كوبرن-جيجر لمواقع التدفق. (الشكل 1). تم استبعاد المناطق التي تحتوي على أقل من خمس محطات تدفق كإجراء احترازي، وتم دمج بعض المناطق بناءً على تصنيفات المناخ الثانوية (على سبيل المثال، المناطق البحر الأبيض المتوسط ذات الصيف الحار والدافئ). كان لكل مجموعة ناتجة 7-13 محطة تدفق تم استخدامها لحساب إحصائيات الدقة.
بشكل عام، كان هناك توافق أفضل بين مجموعة OpenET و ECET في مواقع المحاصيل في المناطق التي تعاني من نقص المياه، من شبه الجاف إلى الجاف (المناطق المتوسطية والصحراوية في الجنوب الغربي) مقارنة بالمناطق الرطبة (الجدول 2 والشكل التوضيحي 2). الري أكثر شيوعًا في المناطق شبه الجافة إلى الجافة، ويميل تبخر المحاصيل إلى أن يكون أقرب إلى معدلات التبخر المحتملة ويتم نمذجته بدقة أكبر في بعض أطر نمذجة RSET. تعتبر الدقة العالية للنماذج في المناطق شبه الجافة والجافة ميزة، نظرًا للأهمية العالية لاستدامة موارد المياه وتحديات الإدارة في هذه المناطق.
من بين المناطق التي تم النظر فيها، كانت قيمة مجموعة OpenET الأكثر دقة لمواقع المحاصيل في المناطق المتوسطية، مع متوسط الخطأ المطلق (MAE) قدره 13.3 و متوسط الجذر التربيعي للخطأ (RMSE) قدره 16.5 مم في الشهر. و من متوسط ECET)، حيث تفوق التجميع على الأعضاء الفرديين. من بين النماذج الفردية، أظهر SIMS أفضل توافق مع ECET في هذه المناطق، مما يشير إلى ظروف ري جيدة لمعظم المواقع أو التأثير المحتمل للمناطق غير المروية المجاورة على نماذج SEB. وبالمثل، في المواقع الجافة (الصحراء الحارة والباردة)، كان لدى SIMS أقل MAE و RMSE (الجدول 2). خلال فترات موسم النمو عندما يتم تطبيق معظم الري، كان NMAE الشهري للتجميع باستمرار أقل من 10% لمواقع الأراضي الزراعية في المناخات المتوسطية (الشكل التوضيحي 2).
كان أداء النموذج في المناطق القارية شبه الرطبة والرطبة في الغرب الأوسط وسهول الوسط مشابهًا لذلك في منطقة المناخ المتوسطي، مرة أخرى مع تفوق المجموعة على النماذج الفردية من حيث الإحصائيات الجماعية (الجدول 2 والشكل التوضيحي 2). كانت الأخطاء أعلى في المواقع الرطبة شبه الاستوائية، حيث كان من الميل أن يبالغ SIMS في تقدير ET مع ميل قدره 1.15 و MBE المعدل من ، مما يشير إلى أن ET أقل ارتباطًا بكثافة الغطاء النباتي في هذه المنطقة، وأن ممارسات الري قد تؤدي إلى إجهاد مائي متقطع للنباتات. تم اقتراح فرضيات لزيادة خطأ RSET في المناطق الرطبة وطرق للتحسين في المناقشة.

الأداء في النظم البيئية الطبيعية

كانت معظم محطات التدفق (61%) المستخدمة في المقارنة بين القياسات في مواقع غير زراعية، بما في ذلك الأراضي الشجرية، والمراعي، والغابات المختلطة، وغابات الصنوبر، والمناطق الرطبة أو المناطق النهرية (الشكل 1). نموذج SIMS
لا يتم تصميمه حاليًا وتنفيذه في أنواع تغطية الأراضي غير الزراعية؛ بالنسبة لهذه البكسلات، يتكون التجميع من خمسة نماذج مع إمكانية إزالة نقطة شاذة واحدة (الطرق). كان خطأ النموذج المنهجي والتباين لمواقع غير الزراعية أعلى من مواقع الأراضي الزراعية (الشكل 5).
أظهرت معظم النماذج انحيازًا عاليًا في مواقع الأراضي الرطبة/الضفاف، حيث هيمنت عليها التنبؤات المفرطة لتبخر الماء خلال الربيع (الشكل 5 من البيانات الموسعة). كان لدى SSEBop دقة أعلى في هذه المواقع مقارنة بالنماذج الأخرى وقيمة التجميع (الجداول التكميلية 2-4). بالنسبة للنماذج التي تقدر جميع مكونات ميزان الطاقة السطحية (DisALEXI وeeMETRIC وgeeSEBAL)، قد ينتج هذا الانحياز عن تقدير منخفض لمصطلح تخزين حرارة الركيزة (الماء) في الربيع قبل أن تتطور مظلة النبات. يمكن تقليل هذه الأخطاء في المستقبل من خلال التصنيف الدقيق للمناطق المغمورة بالمياه.
أظهرت النظم البيئية الطبيعية تحت ضغط مائي عالٍ، مثل الأراضي الشجرية والمراعي في المناخات الصحراوية وشبه الصحراوية في غرب الولايات المتحدة، أعلى مستوى من التباين والخطأ فيما يتعلق بـ ECET (الشكل 5 والجداول التكميلية 2-4). في هذه الأنظمة، يمكن أن تكون ET جزءًا صغيرًا من الطاقة المتاحة، ومن الصعب قياسها على الأرض ونمذجتها باستخدام أساليب RSET. كما أن الأراضي الشجرية تميل إلى أن تكون أكثر تنوعًا من مواقع الأراضي الزراعية، مما يمكن أن يُدخل مزيدًا من عدم اليقين في مقارنات النماذج والقياسات. ومع ذلك، من المهم تقديم تقييم للدقة، سواء لفائدة مراقبة ET وتقييم صحة الأراضي داخل نظم الشجيرات والأراضي العشبية، ولتحديد المجالات الرئيسية للبحث المستقبلي في RSET لتقليل خطأ النموذج.
تطبيقات ET على نطاق لاندسات من OpenET تشمل أيضًا المناظر الطبيعية الغابية، كمؤشر على صحة الغابات ومعدل الوفيات فيها. وكمعيار لاستجابة إنتاج المياه لإدارة الغابات في المواقع الغابية، كانت معظم نماذج OpenET تفرط في تقدير التبخر، لا سيما في مواقع التدفق الدائمة الخضرة التي تم أخذ عينات منها، مما أسفر عن ميل للقيمة الجماعية قدره 1.24 وMBE قدره 16.8 مم في الشهر. في هذه المواقع، أظهر eeMETRIC أقل انحياز مع ميل قدره 1.17 وMBE قدره 10.8 مم في الشهر (17.5%)، بينما بالنسبة لـ MAE وRMSE، تفوق القيمة الجماعية على كل نموذج فردي. ومع ذلك، في مواقع الغابات المختلطة، كان eeMETRIC وDisALEXI في توافق أفضل مع ECET مقارنةً بالتركيبة.

إزالة القيم الشاذة من المجموعة والتباين المكاني بين النماذج

انظر المناقشة التكميلية 2 لتحليل ومناقشة نهج إزالة القيم الشاذة باستخدام MAD الذي يُستخدم لحساب القيمة الجماعية، بما في ذلك التحليل المكاني لحدوث القيم الشاذة والفروق طويلة الأجل بين ET الموسمي لكل نموذج والقيمة الجماعية (الأشكال البيانية للبيانات الموسعة 7 و8، الأشكال التكميلية 3-9 والجداول التكميلية 9 و10). تشير الأدلة إلى أن نهج MAD أظهر مقاييس دقة مشابهة لطرق بسيطة أخرى. على مدار 2016-2022، لم يتم تحديد أي نموذج عادةً كقيمة شاذة في بكسلات الأراضي الزراعية؛ ومع ذلك، كان من المرجح أن يتم تحديد SIMS كقيمة شاذة جماعية بنسبة حوالي 10%، وغالبًا ما أعطى أعلى قيمة ET، لا سيما في السهول الوسطى.

نقاش

يعتبر التبخر والنتح (ET) محركًا حاسمًا ومقياسًا لوظيفة النظام البيئي، والطقس والمناخ، والممارسات الزراعية وإدارة موارد المياه. ومع ذلك، كان من الصعب تقدير ET على نطاق واسع في السابق؛ لذلك، فإن الوصول السهل إلى بيانات ET عالية الدقة (من حيث المكان والزمان) يوفر فوائد اجتماعية لمجموعة متنوعة من أصحاب المصلحة. باستخدام بيانات التبخر الشهري، يمكن لمديري المياه تطوير ميزانيات مائية أكثر دقة لدعم برامج الحفظ المدفوعة بالحوافز واستراتيجيات الإدارة والتجارة المبتكرة. بالنسبة لصانعي السياسات، يمكن أن تحسن هذه البيانات تتبع إمدادات المياه، وتبسط الامتثال التنظيمي، وتعزز تطوير الحلول بالتعاون مع المجتمعات المحلية. قد يتمكن منتجو المحاصيل من تحسين كفاءة ممارسات الري في بعض الحالات، مما يؤدي إلى تعزيز الاستدامة وتقليل
الشكل 3 | المناخ الشهري لنماذج ET المقترنة والملاحظة لمواقع الأراضي الزراعية. أ، المناخ الشهري لنماذج OpenET المقترنة وبرج التدفق من مواقع الأراضي الزراعية. بقايا متوسط ET الشهري (النموذج ناقص متوسط ET المغلق). تشير تسميات غير المغلقة والمغلقة إلى ET لبرج التدفق قبل وبعد تصحيح توازن الطاقة. تمثل الخطوط المتقطعة متوسط ET المغلق زائد خطأين معياريين من المتوسط ومتوسط ET غير المغلق ناقص خطأين معياريين من المتوسط.
تكاليف المياه والأسمدة والطاقة. تستمر المناقشة التكميلية 3 في الحديث عن الحوافز لتحسين كفاءة الري وكيف يمكن لبيانات OpenET أن تقدم قيمة في إطار جدولة الري المعتمد على RSET.
بالإضافة إلى إبلاغ إدارة المياه، فإن OpenET له تطبيقات متعددة في البحث والنمذجة. يمكن أن تستفيد نمذجة الكربون والمناخ من بيانات RSET بدقة 30 مترًا كمؤشر تشخيصي لصحة النظام البيئي واستجابة الوظيفة تحت مناخ متغير. يتم استخدام RSET لتقليل انحياز الحرارة الجافة في فصل الصيف في توقعات الطقس ونماذج المناخ من خلال تحسين تمثيل التبخر من الأراضي المروية. ترابط رطوبة التربة وتقسيم النتح-التبخر يمكن أن تستفيد نماذج الهيدرولوجيا وسطح الأرض على مقاييس متعددة أيضًا من بيانات التبخر والنتح عالية الدقة، على سبيل المثال، كبيانات تحقق أو بيانات قسرية في الأحواض حيث لا تتوفر قياسات تدفق المياه لتقييد ميزانية المياه. .
تحقيق الفوائد الكاملة لبيانات RSET في تطبيقات إدارة الموارد المائية والأراضي يتطلب تقييم دقيق وقابل للتكرار للموثوقية لإبلاغ الممارسين بأفضل ممارسات الاستخدام. تقدم نتائج الدقة التي نقدمها هنا قيودًا قيمة على عدم اليقين في النموذج استنادًا إلى نوع المحصول الواسع، ومنطقة المناخ، والإطار الزمني.
كان متوسط الخطأ في قيمة مجموعة OpenET بالنسبة لمتوسط ECET في مواقع الأراضي الزراعية للـ ET المجمع شهريًا، خلال موسم النمو وسنويًا، يتراوح بين 10% إلى 17% للخطأ المطلق المتوسط و11% إلى 22% لجذر متوسط مربع الخطأ. هذه الأخطاء ضمن مستويات الدقة لـ تم الإبلاغ عن تقنيات الاستشعار عن بُعد تحت الإشراف . وهي تتماشى أيضًا مع أهداف الدقة المحددة من قبل مجموعات مستخدمي OpenET: 10-20% عند
الشكل 4 | متوسط الخطأ المطلق الشهري لمجموعة النماذج لأنواع المحاصيل المختلفة. أ، متوسط التدفق الشهري لبرج القياس . ب، OpenET تجميع MAE و MAE المعدل بواسطة متوسط برج التدفق (NMAE) باستخدام جميع بيانات النموذج المقاسة المزدوجة لمحطات الأراضي الزراعية المجمعة حسب أنواع المحاصيل. لم يتم تضمين المحاصيل السنوية التي كانت لها تاريخ مختلط من التناوب بين أنواع المحاصيل C3 و C4، على سبيل المثال، تناوب الذرة وفول الصويا، في نتائج C3 أو C4 ولكن تم تضمينها في التجميع المشترك.
خطوة زمنية شهرية، و لبيانات ET اليومية . تشمل هذه الأخطاء عدم اليقين في بيانات ECET، والتي يُقدّر أن تتراوح من 10% إلى 30% اعتمادًا على خصائص الموقع وتصميم وصيانة الأدوات. .
قد تدعم هذه النتائج الدقيقة التقدم في تطبيقات إدارة المياه التي تتضمن بيانات OpenET. بالنسبة للأراضي الزراعية، كانت جميع النماذج باستثناء SIMS تحتوي على أخطاء انحياز سلبية عند الخطوة الزمنية الشهرية ( إلى ), مع MBE قدره لقيمة ET المجمعة (MBE لـ SIMS هو ). قد يؤدي الوعي بهذه الأخطاء في الانحياز عند استخدام هذه البيانات لتطبيقات إدارة الري إلى منع الري الناقص غير المقصود الذي يمكن أن يثبط من غلات المحاصيل وإيرادات المزارع. . قد تكون المقارنات المتقاطعة بين نماذج SIMS و PT-JPL المعتمدة بشكل أساسي على الانعكاس والنماذج المدفوعة بـ LST مفيدة لتحديد فترات الإجهاد المائي المتعمد أو غير المتعمد للمحاصيل والري الناقص. تقليل الأخطاء في بيانات OpenET اليومية هو أولوية عالية لتعزيز فائدتها في إدارة المياه في المزارع.
على المقاييس المحلية إلى الإقليمية، يجب أن تُعلم عدم اليقين المبلغ عنه عند الخطوات الزمنية الشهرية إلى السنوية التطبيقات المتعلقة بتوازن المياه، ومحاسبة المياه وإدارة حقوق المياه. يوفر مقارنة بيانات OpenET المجمعة على نطاق مناطق الري أو أحواض المياه مقابل توازنات المياه المقيدة بعناية مسارًا واحدًا لتقييم الانحيازات على مقاييس أكبر. بشكل خاص في إدارة حقوق المياه، يجب الاعتراف بعدم اليقين الحالي في بيانات OpenET (على سبيل المثال، NMAE المجمعة خلال موسم النمو قدره بالنسبة للأراضي الزراعية) عند تقييم استخدام المياه الاستهلاكية، ويجب استخدام بيانات OpenET لهذا الغرض فقط بالاقتران مع مصادر معلومات أخرى.
تقدم هذه الدراسة رؤى حول المسارات المحتملة نحو تحسين دقة النماذج الفردية ضمن مجموعة OpenET. عبر كل من المناظر الطبيعية الزراعية وبعض المناظر الطبيعية الطبيعية، قدّرت معظم النماذج ET للأراضي الزراعية خلال فصل الشتاء والربيع، وخاصة النماذج التي تعتمد على قياسات TIR لحساب ET. قد يكون هذا التقدير المنخفض مرتبطًا بفقدان التباين الحراري عبر الصورة، حيث يتم تقليل الفروق بين أكثر البيكسلات حرارة وبرودة بالنسبة لقيم منتصف الصيف، مما يضيف عدم اليقين إلى أساليب القياس داخل المشهد. قد يكون أيضًا مرتبطًا بسوء تمثيل تبخر التربة خلال فترات الرطوبة الممتدة، وفترات السحاب الممتدة، والخطأ في المدخلات المشتركة للنموذج. بالإضافة إلى ذلك، يمكن أن يؤثر التعامل مع آثار النباتات الراكدة المتقدمة وبقايا المحاصيل على SEB على أداء النموذج خارج موسم النمو. من حيث الأخطاء الملاحظة، فإن خطأ إغلاق توازن الطاقة وعدم اليقين في بيانات EC تتضخم أيضًا خلال الفترات خارج موسم النمو. .
وجدنا زيادة في خطأ النموذج في الأراضي الزراعية في المناخات الرطبة مقارنة بالمناطق الأكثر جفافًا. مرة أخرى، قد تسهم التباينات الحرارية المنخفضة عبر المناظر الطبيعية الرطبة في الأخطاء في نماذج القياس داخل المشهد المعتمدة على TIR. ومع ذلك، فإن المحرك الرئيسي هو على الأرجح الندرة النسبية لاسترجاعات الأقمار الصناعية في السماء الصافية وإمكانية حدوث خطأ في LST بسبب السحب غير المكتشفة. سيكون تحسين أخذ العينات الزمنية لمدخلات نموذج RSET محور تركيز رئيسي في التطوير المستمر في OpenET، من خلال الاستخدام المستقبلي للصور من مستشعرات Landsat-like البصرية (Sentinel-2) والحرارية (ECOSTRESS، VIIRS)، وإدماج الملاحظات المستقبلية لـ TIR من بعثات الأقمار الصناعية التي يتم تطويرها حاليًا بواسطة NASA وUSGS ووكالة الفضاء الأوروبية. يمكن أيضًا تحسين الطرق المستخدمة في حساب قيم ET بين الملاحظات الساتلية الخالية من السحب، والتي تعتمد حاليًا على الاستيفاء الخطي لنسبة ET إلى تدفق مرجعي. تقدم الأساليب المستخدمة في رسم الخرائط وتوقع فينولوجيا النباتات وطرق التحويل الزمني الديناميكي التي تم تطويرها لتطبيقات معالجة الإشارات وعدًا لتقليل الأخطاء الكبيرة خلال فترات التغير السريع في النباتات أو تغطية السحب الممتدة، مما سيساهم في تقليل قيم RMSE عبر مجموعة النماذج.
عند فحص النتائج لفئات المحاصيل المحددة، وجدنا نتائج قوية لـ DisALEXI و SIMS فوق مزارع الكروم، و DisALEXI و SIMS و SSEBop فوق مواقع بساتين الفاكهة والمكسرات – الأهداف الرئيسية لإدارة الري في وادي سنترال. سيساعد زيادة عدد مواقع التحقق في البساتين على معالجة القضايا المتبقية المتعلقة بالنمذجة المرتبطة بهذه البنية المعقدة للسقف. تهدف تجربة استشعار عن بعد لتبخر النتح (T-REX) التي يقودها USDA ARS إلى معالجة هذه الفجوة الملاحظة. .
جميع النماذج، بدرجات متفاوتة، لديها مجال لتحسين ملحوظ في حساب ET في النظم البيئية الطبيعية. على سبيل المثال، تقدر معظم النماذج ET في النظم البيئية الأكثر جفافًا مثل المراعي والأراضي الشجرية بشكل منهجي وتبالغ في تقدير ET في الغابات دائمة الخضرة. قد يؤدي دمج بيانات مرئية وقريبة من الأشعة تحت الحمراء عالية التردد والدقة في نماذج الاستشعار عن بعد إلى تحسين قدرتها على التقاط التحولات الفينولوجية، خاصة في المناطق الجافة/شبه الجافة، والأنظمة الزراعية بشكل عام. . قد يؤدي تحسين مدخلات نماذج الأرصاد الجوية الموزعة , وبيانات تصنيف استخدام الأراضي وبيانات التربة إلى تحسين أداء النموذج في كل من النظم البيئية الطبيعية وفي الأراضي الزراعية. بشكل خاص، يمكن إعادة تقييم مجموعات البيانات المجمعة من محطات الأرصاد الجوية الزراعية المستخدمة لحساب أسطح تصحيح الانحياز لـ ET المرجعي لضمان توافق السطح المرجعي مع افتراضات معادلة بنمان-مونتيث لجمعية المهندسين المدنيين الأمريكية. .
ستستهدف تقييمات دقة OpenET المستقبلية الأسباب الرئيسية للأخطاء في قياسات ET الأرضية وطرق RSET. تشمل العوامل المحددة التي يجب مراعاتها التأثيرات المحلية على ET المحسوبة والمقاسة، وإغلاق ميزانية الطاقة EC، والتباين الحراري المحلي، وتقليل ET في الأنظمة المروية الناقصة أو المعتمدة على الأمطار، والانحيازات المحتملة في المدخلات الجوية الموزعة إلى نماذج RSET، والتقاط دقيق لـ ET عبر المناظر الطبيعية المزروعة بشكل نادر. المقارنات مع أخرى
الشكل 5 | مجموعة النماذج الشهرية مقابل ET المرصود للمواقع المجمعة حسب نوع استخدام الأراضي. مقارنة شهرية لـ OpenET الشهرية مقابل برج التدفق المغلق لكل مجموعة عامة من استخدام الأراضي. تم تضمين نتيجة نموذج الانحدار الخطي الأقل تربيعًا لكل مجموعة و .
قد توفر منتجات ET الموزعة مكانيًا مثل MOD16 أو FLUXCOM رؤى إضافية لرسم خرائط ET العالمية التشغيلية على مقاييس الحقول ( ). ستساعد المقارنات ضد بيانات ET المحسوبة من دراسات توازن المياه طويلة الأجل في سد الفجوات في التغطية المكانية في ET المقاس في الموقع عبر غرب الولايات المتحدة في المناطق الهيدرولوجية المهمة ولكن المزروعة بشكل نادر مثل حوض نهر كولورادو العلوي.

الاستنتاجات

توفر منصة OpenET بيانات ET مستمرة مكانيًا عند القرار في جميع أنحاء غرب الولايات المتحدة. شملت مقارنة وتقييم دقة ستة نماذج RSET المعتمدة على الأقمار الصناعية التي تشكل النسخة الحالية من OpenET، وET الجماعي المحسوب من النماذج الستة، ومجموعة بيانات تدفق الإدي المعروفة جيدًا من 152 محطة تقع في الولايات المتحدة المتجاورة. استنادًا إلى نتائج من 59 محطة ET للمحاصيل الزراعية تقع في مجموعة متنوعة من المناطق المناخية، لم يُلاحظ أي انحياز منهجي للنموذج في الأراضي الزراعية، وكانت مقاييس الخطأ ضمن أو قريبة من الأهداف التي وضعتها شركاء OpenET بما في ذلك المزارعين ومديري الري ووكالات إدارة المياه. كانت أفضل مقاييس الدقة مرتبطة بالمقاييس الزمنية الموسمية والسنوية، وللمحاصيل في المناطق الجافة/شبه الجافة. عادةً ما كان متوسط OpenET الجماعي، مع إزالة القيم الشاذة، يتفوق على أي نموذج فردي من حيث إحصائيات الخطأ. بشكل عام، لم يتم تحديد أكثر من نموذج واحد كقيمة شاذة خلال أشهر موسم النمو في معظم المناطق الزراعية في غرب الولايات المتحدة، وغالبًا ما لم يتم استبعاد أي نماذج. تسلط هذه النتيجة الضوء على التقدم الكبير الذي تم تحقيقه حتى الآن في تطوير أساليب نمذجة RSET المؤتمتة بالكامل التي يمكن استخدامها لرسم ET على مساحات واسعة بدقة على مستوى الحقل. حددت الدراسة مسارات للبحث المستهدف المستقبلي وتحسين النماذج، وتهدف إلى دعم مجتمع بحث RSET في تطوير تقنيات RSET أكثر قوة ودقة. نحن أيضًا نأمل أن يوفر هذا التقييم ثقة إضافية لمديري موارد المياه، والمزارعين، ورعاة الماشية، والعلماء وغيرهم.
المستخدمون المحتملون لـ OpenET بسبب الصرامة العالية والشفافية في الأساليب التي تم استخدامها.

طرق

معالجة بيانات التدفق وأخذ عينات البصمة

استخدمنا مجموعة بيانات ET المستندة إلى تدفق إيدي تم تنسيقها بعناية وأدوات لاستخدامها في هذا التقييم والتقييمات اللاحقة لنماذج OpenET RSET تُوصف الأسس وخطوات اتخاذ القرار لجمع ومعالجة بيانات التدفق، بالإضافة إلى تحليلات تقنيات أخذ العينات من البصمة وخطأ إغلاق توازن الطاقة ضمن مجموعة البيانات، في دراسة فولك وآخرون. تمت معالجة تقنيات معالجة البيانات لملء الفجوات وتصحيح أخطاء إغلاق توازن الطاقة باستخدام أدوات بايثون مفتوحة المصدر. التي تعزز من أصل البيانات وقابليتها للتكرار. كما خضعت البيانات لعمليات فحص وتصفية قائمة على الجودة والرؤية. تتكون مجموعة البيانات النهائية المعالجة من 161 محطة، وهي عامة وتشتمل على بيانات ET اليومية والشهرية وبيانات الأرصاد الجوية، ورسوم بيانية تفاعلية لهذه البيانات لكل محطة، ومعلومات الموقع مثل استخدام الأراضي واعترافات الباحث الرئيسي. نلاحظ أن تسع محطات في مجموعة البيانات لم تُدرج في النتائج الإحصائية المقدمة هنا لأنها كانت تحتوي على تغطية بيانات لا تتداخل مع البيانات التي يمكن تطويرها لجميع نماذج OpenET الستة. على سبيل المثال، لم يكن من الممكن تنفيذ جميع النماذج من صور الأقمار الصناعية المسجلة قبل عام 2001 (المرجع 5). تُظهر الشكل 1 خريطة للمحطات الـ 152 المستخدمة في هذا التقييم الدقيق بالإضافة إلى أنواع تغطية الأراضي والمناطق المناخية وفقًا لكوبن-جيجر، وتوفر الجدول التكميلية 1 بيانات وصفية إضافية لكل محطة.
تم تنزيل بيانات الغالبية (106) من محطات التدفق في هذه الدراسة من موقع AmeriFlux، الذي تم الوصول إليه آخر مرة في 27 أكتوبر 2020، وتم استرجاع المحطات المتبقية من مجموعة متنوعة من المصادر والمحققين الرئيسيين من الشركاء الجامعيين، والمسح الجيولوجي الأمريكي، ووزارة الزراعة الأمريكية وغيرها. . بالإضافة إلى أنظمة EC، هناك أربعة ميزان دقيق لقياس ET للأراضي الزراعية في تكساس وسبعة نسب بوان عالية الجودة
المواقع المجهزة، التي تقيس ET في الغالب في أراضي الشجيرات الفرياطية في نيفادا ، تم تضمينها في مجموعة البيانات. تم ملء الفجوات في التدفقات الأولية نصف الساعية لأربعة مكونات رئيسية لتوازن الطاقة – التدفق الكامن، التدفق الحساس، وتدفق حرارة التربة، والإشعاع الصافي – باستخدام الاستيفاء الخطي حيث تم استيفاء الفجوات التي تصل إلى ساعتين خلال النهار أو أربع ساعات خلال الليل. إذا كانت إذا كانت الفترة لا تزال تحتوي على فجوات، فلم يتم حساب المتوسط اليومي وترك قيمة التدفق اليومي كفجوة. بعد ملء هذه الفجوات الأولية، تم حساب متوسطات التدفقات لفترات يومية وتم تطبيق تصحيح إغلاق توازن الطاقة وفقًا لنهج نسبة توازن الطاقة اليومية المحدد بواسطة FLUXNET2015/ONEFlux. تم استخدام تدفق الحرارة الكامنة اليومي المصحح، وهو الطاقة المستهلكة من خلال التبخر، لحساب التبخر مع تعديل لحرارة تبخر الماء لدرجة حرارة الهواء. . تُعرف هذه القيمة المعدلة للإغلاق باسم تدفق مغلق ET أو ET المقاس في النص الرئيسي، وجميع القياسات الإحصائية المبلغ عنها لنماذج OpenET كانت مقابل بيانات ET المصححة بتوازن الطاقة. تم ملء فجوات ET اليومية لاحقًا باستخدام نسبة ET المرجعي من gridMET ومرجع العشب من gridMET. لاستبعاد محطات التدفق ذات عدم اليقين العالي في البيانات، تم اختيار فقط المحطات التي لديها متوسط إغلاق توازن الطاقة اليومي 0.75 أو أعلى خلال موسم النمو و0.6 أو أعلى خلال موسم غير النمو لهذه المقارنة. هنا، تم رسم فترات موسم النمو بشكل مكاني استنادًا إلى نهج درجات الحرارة التراكمية وصدمة الصقيع المستمدة من بيانات المناخ الموزعة على المدى الطويل وهي محددة لكل موقع تدفق. تتشابه مجموعة البيانات النهائية مع إصدار FLUXNET2015 الأخير (المرجع 29) الذي يتكون من بيانات محطات تدفق الإدي عالية الجودة التي خضعت لعمليات معالجة وتصحيح مماثلة. أكبر اختلاف بين مجموعتي البيانات، من حيث تقديرات تدفق الحرارة الكامنة اليومية، ينجم عن إجراءات ملء الفجوات المختلفة، حيث تعتبر طريقتنا أبسط وأكثر تحفظًا. .
تم استخدام نهجين لتقدير بصمات أبراج التدفق أو منطقة المصدر لعينات بكسل البرج من صور RSET: (1) شبكات بكسل ‘ثابتة’ مربعة بسيطة (30 م من لاندسات) من و محددة حول مواقع المحطات، و(2) تقديرات منطقة مصدر التدفق المستندة إلى نموذج ثنائي الأبعاد، تم نمذجتها باستخدام بيانات الأرصاد الجوية الساعية باستخدام Kljun وآخرون. النهج، مع تحويل بصمات الساعة إلى بصمات يومية/شهرية متوسطة مرجحة بواسطة المرجع . تم تحديد موضع الشبكات الثابتة استنادًا إلى صور عالية الدقة لتجنب تضمين بكسلات تغطية أرض غير تمثيلية (الهياكل، الطرق والقنوات)، وتم تحريكها قليلاً في اتجاه الرياح السائد كما تحدده متوسطات الرياح النهارية على المدى الطويل (المبنية على بيانات بين الساعة 6:00 و20:00 بالتوقيت المحلي). على الرغم من أن البصمات المعتمدة على الفيزياء والديناميكية الزمنية كانت مفضلة على البصمات الثابتة، إلا أن حوالي نصف المحطات في مجموعة البيانات كانت تحتوي على بيانات كافية لإنتاجها. عادةً، كان هناك واحد أو أكثر من معلمات الإدخال إلى Kljun et al. نموذج، مثل الانحراف المعياري لمكون الرياح العرضية الناتج عن الاضطراب أو سرعة الاحتكاك، لم يكن متاحًا. وصف تفصيلي لتقدير المعلمات، وخطوات المعالجة، والطريقة المستخدمة لإنشاء صور متوسط الأثر الموزون (باستخدام ET المرجعي من بيانات الطقس الموزعة NLDAS2). يمكن العثور عليه في فولك وآخرون. . كما أجرينا مقارنة دقيقة بين تقاطعات مناطق المصدر من الشبكات الثابتة بأحجام مختلفة والأثر الزمني الديناميكي. وكانت النتيجة الرئيسية هي أن الأكبر كانت الشبكات تميل إلى تضمين مساحة البصمة المحددة ديناميكيًا بشكل أكبر بكثير من الأحجام الأصغر للشبكة في المتوسط؛ ومع ذلك، كانت الأحجام الأصغر كانت الشبكات تميل إلى التداخل مع البكسلات التي اعتُبرت جزءًا من البصمة الديناميكية بشكل أكثر اتساقًا. لذلك، قررنا استخدام الشبكات لأخذ عينات البكسل في معظم مواقع التدفق حيث لم يكن من الممكن إنشاء بصمة ديناميكية، مع استثناءات للمواقع ذات المحيطات غير المتجانسة أو مع تغطية أرضية غير تمثيلية بالقرب من المحطة. بالنسبة لهذه المواقع، استخدمنا أو الشبكات لتجنب إعطاء وزن متساوٍ للبكسلات التي قد تحتوي على تغطية أرضية مختلفة والتي تقع بالقرب من محيط منطقة الأثر الفعلي النموذجية .

بيانات النموذج

تستند الغالبية العظمى من النماذج التي تشكل مجموعة OpenET إلى تنفيذات كاملة أو مبسطة من نهج SEB. يأخذ نهج SEB في الاعتبار الطاقة المستخدمة لتحويل الماء السائل في النباتات والتربة إلى بخار يتم إطلاقه في الغلاف الجوي. يعتمد نهج SEB على قياسات الأقمار الصناعية لدرجة حرارة السطح وانعكاسية السطح، جنبًا إلى جنب مع متغيرات رئيسية أخرى لسطح الأرض والطقس، لحساب مكونات توازن الطاقة – الإشعاع الصافي، تدفق الحرارة الحسية، تدفق الحرارة الأرضية وتدفق الحرارة الكامنة. eeMETRIC جي سيبال وديساليكسي احسب كل مكون من مكونات ميزان الطاقة باستخدام البيانات البصرية (أي، الموجات القصيرة) والحرارية (أي، الموجات الطويلة)، بينما SSEBop و PT-JPL هي طرق مبسطة حيث لا يتم حساب بعض مكونات ميزان الطاقة، أو يتم حسابها باستخدام مجموعة من الافتراضات المبسطة. SIMS يعتمد على بيانات انعكاس السطح، ومعلومات نوع المحاصيل، ونموذج توازن المياه في التربة الموزعة على شبكة لحساب التبخر والنتح كدالة لكثافة السطح باستخدام نهج معامل المحصول للأراضي الزراعية.
محرك جوجل إيرث تم استخدام واجهة برمجة التطبيقات الخاصة بلغة بايثون لتطوير سير عمل لأخذ عينات من بيانات نموذج RSET في OpenET في مواقع تدفق ET. تم أخذ عينات من بيانات نموذج RSET اليومية والشهرية في كل موقع باستخدام مجموعة من الثوابت. و/أو ) ومناطق مصادر التدفق الديناميكي. كانت الشروط لكل من طرق الاستخراج باستخدام بصمات ثابتة كما يلي: (1) تم حساب ET اليومي من eeMETRIC وSIMS وSSEBop للمواقع خارج كاليفورنيا كحاصل ضرب متوسط الكسر اليومي من ET المرجعي للعشب (EToF) الذي تنتجه النماذج ومتوسط ET المرجعي للعشب المصحح بالتحيز من gridMET (ETo) (تم تكرار ذلك للمواقع داخل كاليفورنيا باستخدام ETo اليومي من CIMIS، حيث يتم استخدام CIMIS بشكل أكثر شيوعًا والاعتماد عليه في كاليفورنيا)؛ (2) تم حساب ET اليومي من PT-JPL وgeeSEBAL وALEXI/DisALEXI لجميع المواقع كمتوسط مكاني لبكسلات ET اليومية التي تنتجها النماذج؛ (3) تم حساب ET الشهري من جميع نماذج RSET للمواقع خارج كاليفورنيا كحاصل ضرب متوسط EToF الشهري ومتوسط ETo من gridMET (تم تكرار ذلك للمواقع داخل كاليفورنيا باستخدام ETo الشهري من CIMIS). عملية استقراء البيانات الفورية (وقت المرور) إلى ET اليومي هي حساب داخلي للنموذج وتختلف لكل نموذج، ونشير إلى القراء إلى الوثائق الفردية للنماذج للحصول على التفاصيل بالإضافة إلى Melton et al. يتم وضع علامة على بكسلات صور لاندسات اليومية الملوثة بالسحب بناءً على مؤشرات مستمدة من CFMask. في نطاق ضمان جودة البكسل (QA_PIXEL) ولا تُعتبر تلك البكسلات. عند حساب ET الشهري، يتم حساب جميع بكسلات ET اليومية المفقودة أو المظللة عن طريق الاستيفاء الخطي بين أقرب بكسلات غير مظللة (خالية من السحب) زمنياً ضمن أيام.
كانت شروط كل من طرق الاستخراج باستخدام بصمات ديناميكية كما يلي:
(1) تم حساب ET اليومي من eeMETRIC وSIMS وSSEBop للمواقع خارج كاليفورنيا من خلال ضرب وحدات EToF اليومية المأخوذة من النماذج في الوزن اليومي لكل وحدة تدفق للحصول على وحدات EToF اليومية الموزونة، وجمع جميع وحدات EToF اليومية الموزونة للحصول على متوسط EToF اليومي الموزون، وتطبيع متوسط EToF اليومي الموزون من خلال مجموع الأوزان لمراعاة الأوقات التي لم يساوي فيها مجموع الأوزان 1 (على سبيل المثال، بسبب حجب السحب لوحدات البكسل)، ثم ضرب متوسط EToF اليومي الموزون في متوسط ETo اليومي المصحح من bias من gridMET (الذي تم استبداله للمواقع داخل كاليفورنيا باستخدام ETo اليومي من CIMIS);
(2) تم حساب ET اليومي من PT-JPL و geeSEBAL و ALEXI/DisALEXI لجميع المواقع عن طريق ضرب بكسلات ET اليومية في أوزان بصمة التدفق اليومية للحصول على بكسلات ET اليومية الموزونة، وجمع جميع بكسلات ET اليومية الموزونة للحصول على ET اليومي الموزون المتوسط، ثم تطبيع ET اليومي الموزون المتوسط بواسطة مجموع الأوزان، و
(3) تم حساب ET الشهري من جميع نماذج RSET للمواقع خارج كاليفورنيا من خلال ضرب بكسلات EToF الشهرية بـ
تُستخدم أوزان بصمة التدفق الشهرية للحصول على بكسلات EToF الشهرية الموزونة، من خلال جمع جميع بكسلات EToF الشهرية الموزونة للحصول على متوسط EToF الشهري الموزون، ثم يتم تطبيع متوسط EToF الشهري الموزون بواسطة مجموع الأوزان، وبعد ذلك يتم ضرب متوسط EToF الشهري الموزون في متوسط ETo المصحح بالتحيز من شبكة gridMET (الذي تم استبداله للمواقع داخل كاليفورنيا باستخدام ETo الشهري من CIMIS).
كان من الضروري إجراء معالجة إضافية بعد استخراج ET اليومي عندما تم استخراج أيام بيانات مكررة في مواقع مختارة بسبب تداخل مسارات لاندسات. في بعض الأحيان، كان الموقع يقع ضمن بصمات مشهدين متداخلين من لاندسات، مما أدى إلى وجود أكثر من قيمة ET في تاريخ مرور معين. للحصول على قيم ET يومية فردية للموقع، تم حساب المتوسط المرجح اليومي لـ ET لكل يوم باستخدام عدد البكسلات (أي عدد البكسلات المستخدمة عند اشتقاق قيمة ET المكانية المتوسطة المعنية) كوزن. كانت أعداد بكسلات ET أحيانًا أقل من إجمالي الشبكة/البصمة بسبب إزالة البكسلات ذات الجودة الرديئة (على سبيل المثال، حجب السحب).

حساب جماعي

تم حساب المتوسط الجماعي للنماذج الستة من OpenET بعد إزالة ما يصل إلى نموذجين شاذين بناءً على MAD. ، مقياس قوي للتشتت مناسب للعينات الصغيرة. يتم إزالة القيم الشاذة على مستوى البكسل لكل صورة ET تم إنشاؤها. لتحديد القيم الشاذة لمشهد واحد، يتم أولاً حساب القيمة الوسيطة وMAD من الوسيط كالتالي
أين هو قيمة ET للنموذج و هو المجموعة الكاملة من تقديرات ET لجميع النماذج الستة. هنا، هو عدد قياسي محدد بـ 1.483، وقد تم اشتقاقه بناءً على فرضية طبيعية عينة السكان. . يُشار إلى هذا النهج أحيانًا بقانون MADe، حيث عادةً ما يتم تعديل قيمة MAD بواسطة أو 3 بناءً على تقييم ذاتي للبيانات، والتي تُستخدم بعد ذلك لإنشاء نطاق حول الوسيط:
تُعتبر تقديرات النموذج التي تقع خارج النطاق كقيم شاذة، ويتم إزالة ما يصل إلى قيمتين شاذتين (الأبعد عن الوسيط) من مجموعة تقديرات النموذج قبل أخذ المتوسط الجماعي.
نظرًا لميول بعض نماذج OpenET للتنبؤ بمعدل ET صفر أو حتى معدلات ET سلبية في بعض المناطق الجافة خلال الفترات الجافة، قمنا بتعديل النهج المذكور أعلاه لهذه السيناريوهات. على وجه التحديد، عندما يكون تقدير الوسيط الجماعي صفرًا ولكن على الأقل نموذج واحد يتنبأ بمعدل ET إيجابي، يتم أخذ المتوسط الجماعي ليشمل تلك القيمة دون أي إزالة مسبقة للقيم الشاذة. في هذه الحالات، ستؤدي إزالة القيم الشاذة إلى إزالة تقديرات النماذج التي تكون إيجابية، وعلى الرغم من أن ET الفعلي قد يكون ضئيلًا جدًا، إلا أن تقدير الصفر لا يعتبر واقعيًا من الناحية الفيزيائية. ومع ذلك، في هذه السيناريوهات، نظرًا لأن الغالبية العظمى من النماذج قد تتنبأ بالصفر، فإن المتوسط الجماعي سيكون أيضًا مائلًا بشدة نحو الصفر مما يجعل هذا إجراءً محافظًا لمنع تقديرات جماعية صفرية.

التحليلات الإحصائية

إحصائيات ملخصة رئيسية بما في ذلك ميل الانحدار الخطي باستخدام المربعات الصغرى المفروض أن يمر عبر الأصل (الميل) بالإضافة إلى الانحدار الخطي مع تقاطع (الجدول التكميلي 7)، MBE، MAE، RMSE ومعامل التحديد ( تم حساب ( ) باستخدام ملاحظات متزاوجة بين تقديرات ET لنموذج OpenET وتقديرات ET للفلوق المصححة والمعالجة بعد. لم تتم مقارنة إحصائيات الدقة اليومية مع أي بيانات ET لمحطات مملوءة بالفجوات، واستخدمت الإحصائيات الشهرية فقط بيانات ET لمحطات بها 5 أيام أو أقل مملوءة بالفجوات في الشهر. استخدمت تقييمات موسم النمو والتقييمات السنوية بيانات شهرية متزاوجة ولم تشمل أي فترات بها فجوات شهرية. أيضًا، العدد
كانت ملاحظات الأزواج دائمًا هي نفسها بين النماذج لجميع التحليلات الإحصائية.
تم حساب جميع الإحصائيات على أساس كل موقع باستخدام ET المقاس بالنموذج المقترن باستخدام حزمة Numpy في بايثون الإصدار 1.17.2 (المرجع 63). بالنسبة للانحدار الخطي، تم استخدام خوارزمية Numpy linalg.Istsq، والتي تطبق طريقة المربعات الصغرى. استخدمنا ET النموذجي كمتغير تابع وET المقاس كمتغير مستقل.
تم حساب MBE على أنه
أين هو الملاحظ هو ET المتوقع من النموذج و هو العدد الإجمالي لنقاط بيانات ET المقاسة بواسطة النموذج المزدوج.
تم حساب MAE على أنه
وتم حساب RMSE على أنه
هنا، تم حساب القيم ك مربع معامل ارتباط بيرسون، الذي تم حسابه من بيانات ET المقاسة والمودلة المزدوجة باستخدام حزمة statsmodels في بايثون، الإصدار 0.12.1 (المرجع 64).
لجمع الإحصائيات حسب نوع الغطاء الأرضي أو منطقة المناخ، استخدمنا طريقتين: (1) لحساب الانحدار الخطي و تم تجميع جميع البيانات من كل ملاحظة أرضية في مجموعة (على سبيل المثال، تقديرات ET الشهرية المزدوجة لنموذج المحطة لمحطات المحاصيل السنوية) معًا قبل حساب إحصائية واحدة لكل نموذج؛ و(2) تم حساب MBE وMAE وRMSE بشكل منفصل لكل محطة أرضية، ثم تم أخذ متوسط مرجح. كانت الإحصائيات المجمعة مرجحة بجذر عدد الملاحظات المزدوجة لكل محطة. السبب هو تجنب إعطاء وزن كبير لمحطات ذات سجلات بيانات طويلة بشكل مفرط، مع عدم إعطاء وزن متساوٍ أيضًا لمحطات ذات سجلات بيانات قصيرة. كما فرضنا متطلبات طول البيانات لمحطات ET في الموقع: لكي يتم تضمينها في إحصائيات المتوسطات اليومية المجمعة، طلبنا من المحطات أن تحتوي على حد أدنى من ست نقاط بيانات مرتبطة بين المحطة والنموذج، وحد أدنى من ثلاث ملاحظات مرتبطة لتضمينها في إحصائيات المتوسطات الشهرية المجمعة. نلاحظ أن ميلتون وآخرون قدمت مقاييس إحصائية مماثلة من مجموعة فرعية من مواقع الأراضي الزراعية المستخدمة في هذه الدراسة، وفي تلك الدراسة، كان ميل الانحدار الخطي و المقاييس قد شملت الوزن، الذي اعتبرناه غير مناسب أو غير ضروري في هذه الدراسة. من أجل التوافق، تم حساب الإحصائيات بنفس الطريقة كما في ميلتون وآخرون. مقدمة في الجدول التكميلي 12.
تم استخدام اختبار توكي بعد الحدث، المعروف أيضًا باختبار الفرق المعنوي الصادق، لمقارنة تقديرات متوسط ET المتعددة من كل نموذج، والمتوسط الجماعي، ومن متوسط بيانات تدفق ET المغلقة وغير المغلقة. تم تطبيق الاختبار باستخدام جميع البيانات المزدوجة من محطات الأراضي الزراعية، بما في ذلك لمجموعات المحاصيل الفرعية: المحاصيل السنوية، والبساتين، وكروم العنب، على مقاييس زمنية يومية، شهرية، خلال موسم النمو وسنوية. تم تعيين معدل الخطأ العائلي إلى 0.05 وتم إجراء الاختبار باستخدام حزمة statsmodels في بايثون، الإصدار 0.12.1 (المرجع 64).

ملخص التقرير

معلومات إضافية حول تصميم البحث متاحة في ملخص تقارير مجموعة ناتشر المرتبط بهذه المقالة.

توفر البيانات

بيانات ET المقاسة في الموقع التي تم تحليلها خلال الدراسة الحالية متاحة في مستودع زينودو، مع المعرفhttps://doi. org/10.5281/zenodo.7636781. تم تحليل بيانات نموذج OpenET ET
خلال الدراسة الحالية متاحة في مستودع زينودو، مع المعرفhttps://doi.org/10.5281/zenodo.10119477.

توفر الشيفرة

الشفرة المستخدمة لمعالجة بيانات برج تدفق الدوامة للدراسة الحالية متاحة للجمهور على GitHub (https://github.com/Open-ET/flux-data-qaqc). الكود المستخدم لإنشاء بصمات التدفق للدراسة الحالية متاح للجمهور على GitHub (https://github.com/Open-ET/تدفق-بيانات-البصمة).

References

  1. Fisher, J. B. et al. The future of evapotranspiration: global requirements for ecosystem functioning, carbon and climate feedbacks, agricultural management, and water resources. Water Resour. Res. 53, 2618-2626 (2017).
  2. Dieter, C. A. et al. Estimated use of water in the United States in 2015. Circular 1411 https://pubs.usgs.gov/publication/cir1441 (2018).
  3. Cook, B. I., Ault, T. R. & Smerdon, J. E. Unprecedented 21st century drought risk in the American Southwest and Central Plains. Sci. Adv. 1, e1400082 (2015).
  4. Liu, P.-W. et al. Groundwater depletion in California’s Central Valley accelerates during megadrought. Nat. Commun. 13, 7825 (2022).
  5. Melton, F. S. et al. OpenET: filling a critical data gap in water management for the western United States. J. Am. Water Resour. Assoc. 58, 971-994 (2022).
  6. Chen, J. M. & Liu, J. Evolution of evapotranspiration models using thermal and shortwave remote sensing data. Remote Sens. Environ. 237, 111594 (2020).
  7. Anderson, M. et al. Field-scale assessment of land and water use change over the California Delta using remote sensing. Remote Sens. 10, 889 (2018).
  8. Allen, R. G., Tasumi, M. & Trezza, R. Satellite-based energy balance for mapping evapotranspiration with internalized calibration (METRIC)-Model. J. Irrig. Drain. Eng. 133, 380-394 (2007).
  9. Laipelt, L. et al. Long-term monitoring of evapotranspiration using the SEBAL algorithm and Google Earth Engine cloud computing. ISPRS J. Photogramm. Remote Sens. 178, 81-96 (2021).
  10. Fisher, J. B., Tu, K. P. & Baldocchi, D. D. Global estimates of the land-atmosphere water flux based on monthly AVHRR and ISLSCP-II data, validated at 16 FLUXNET sites. Remote Sens. Environ. 112, 901-919 (2008).
  11. Pereira, L. S. et al. Prediction of crop coefficients from fraction of ground cover and height. Background and validation using ground and remote sensing data. Agric. Water Manag. 241, 106197 (2020).
  12. Melton, F. S. et al. Satellite irrigation management support with the terrestrial observation and prediction system: a framework for integration of satellite and surface observations to support improvements in agricultural water resource management. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 5, 1709-1721 (2012).
  13. Senay, G. B. et al. Improving the operational simplified surface energy balance evapotranspiration model using the forcing and normalizing operation. Remote Sens. 15, 260 (2023).
  14. Gorelick, N. et al. Google Earth Engine: planetary-scale geospatial analysis for everyone. Remote Sens. Environ. 202, 18-27 (2017).
  15. Allen, R. G. et al. Satellite-based energy balance for mapping evapotranspiration with internalized calibration (METRIC)Applications. J. Irrig. Drain. Eng. 133, 395-406 (2007).
  16. Knipper, K. R. et al. Using high-spatiotemporal thermal satellite ET retrievals for operational water use and stress monitoring in a California vineyard. Remote Sens. 11, 2124 (2019).
  17. Senay, G. B., Friedrichs, M., Singh, R. K. & Velpuri, N. M. Evaluating Landsat 8 evapotranspiration for water use mapping in the Colorado River Basin. Remote Sens. Environ. 185, 171-185 (2016).
  18. Foster, T., Mieno, T. & Brozović, N. Satellite-based monitoring of irrigation water use: assessing measurement errors and their implications for agricultural water management policy. Water Resour. Res. 56, e2020WRO28378 (2020).
  19. Volk, J. M. et al. Development of a benchmark eddy flux evapotranspiration dataset for evaluation of satellite-driven evapotranspiration models over the CONUS. Agric. For. Meteorol. 331, 109307 (2023).
  20. Volk, J. M. et al. Post-processed data and graphical tools for a CONUS-wide eddy flux evapotranspiration dataset. Data Brief https://doi.org/10.1016/j.dib.2023.109274 (2023).
  21. Baldocchi, D. Measuring fluxes of trace gases and energy between ecosystems and the atmosphere-the state and future of the eddy covariance method. Glob. Change Biol. 20, 3600-3609 (2014).
  22. Baldocchi, D. et al. FLUXNET: a new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide, water vapor, and energy flux densities. Bull. Am. Meteorol. Soc. 82, 2415-2434 (2001).
  23. Hampel, F. R. The influence curve and its role in robust estimation. J. Am. Stat. Assoc. 69, 383-393 (1974).
  24. Leys, C., Ley, C., Klein, O., Bernard, P. & Licata, L. Detecting outliers: do not use standard deviation around the mean, use absolute deviation around the median. J. Exp. Soc. Psychol. 49, 764-766 (2013).
  25. Thompson, P. D. How to improve accuracy by combining independent forecasts. Mon. Weather Rev. 105, 228-229 (1977).
  26. Kirtman, B. P. et al. The North American multimodel ensemble: phase-1 seasonal-to-interannual prediction; phase-2 toward developing intraseasonal prediction. Bull. Am. Meteorol. Soc. 95, 585-601 (2014).
  27. Bai, Y. et al. On the use of machine learning based ensemble approaches to improve evapotranspiration estimates from croplands across a wide environmental gradient. Agric. For. Meteorol. 298, 108308 (2021).
  28. Novick, K. A. et al. The AmeriFlux network: a coalition of the willing. Agric. For. Meteorol. 249, 444-456 (2018).
  29. Pastorello, G. et al. The FLUXNET2O15 dataset and the ONEFlux processing pipeline for eddy covariance data. Sci. Data 7, 1-27 (2020).
  30. Mauder, M., Foken, T. & Cuxart, J. Surface-energy-balance closure over land: a review. Bound. Layer Meteorol. 177, 395-426 (2020).
  31. Ingwersen, J., Imukova, K., Högy, P. & Streck, T. On the use of the post-closure methods uncertainty band to evaluate the performance of land surface models against eddy covariance flux data. Biogeosciences 12, 2311-2326 (2015).
  32. Knipper, K. R. et al. Evapotranspiration estimates derived using thermal-based satellite remote sensing and data fusion for irrigation management in California vineyards. Irrig. Sci. 37, 431-449 (2019).
  33. Bambach, N. et al. Evapotranspiration uncertainty at micrometeorological scales: the impact of the eddy covariance energy imbalance and correction methods. Irrig. Sci. 40, 445-461 (2022).
  34. Rubel, F., Brugger, K., Haslinger, K. & Auer, I. The climate of the European Alps: shift of very high resolution Köppen-Geiger climate zones 1800-2100. Meteorol. Z. 26, 115-125 (2017).
  35. Yang, Y. et al. Studying drought-induced forest mortality using high spatiotemporal resolution evapotranspiration data from thermal satellite imaging. Remote Sens. Environ. 265, 112640 (2021).
  36. Isaacson, B. N., Yang, Y., Anderson, M. C., Clark, K. L. & Grabosky, J. C. The effects of forest composition and management on evapotranspiration in the New Jersey pinelands. Agric. For. Meteorol. 339, 109588 (2023).
  37. Qian, Y. et al. Neglecting irrigation contributes to the simulated summertime warm-and-dry bias in the central United States. Npj Clim. Atmos. Sci. 3, 31 (2020).
  38. Lei, F., Crow, W. T., Holmes, T. R., Hain, C. & Anderson, M. C. Global investigation of soil moisture and latent heat flux coupling strength. Water Resour. Res. 54, 8196-8215 (2018).
  39. Dong, J., Lei, F. & Crow, W. T. Land transpiration-evaporation partitioning errors responsible for modeled summertime warm bias in the central United States. Nat. Commun. 13, 336 (2022).
  40. Abolafia-Rosenzweig, R., Pan, M., Zeng, J. & Livneh, B. Remotely sensed ensembles of the terrestrial water budget over major global river basins: an assessment of three closure techniques. Remote Sens. Environ. 252, 112191 (2021).
  41. Wang, Q. et al. Land surface models significantly underestimate the impact of land-use changes on global evapotranspiration. Environ. Res. Lett. 16, 124047 (2021).
  42. Allen, R. G., Pereira, L. S., Howell, T. A. & Jensen, M. E. Evapotranspiration information reporting: I. Factors governing measurement accuracy. Agric. Water Manag. 98, 899-920 (2011).
  43. Adu, M. O., Yawson, D. O., Armah, F. A., Asare, P. A. & Frimpong, K. A. Meta-analysis of crop yields of full, deficit, and partial root-zone drying irrigation. Agric. Water Manag. 197, 79-90 (2018).
  44. Xue, J. et al. Improving the spatiotemporal resolution of remotely sensed ET information for water management through Landsat, Sentinel-2, ECOSTRESS and VIIRS data fusion. Irrig. Sci. 40, 609-634 (2022).
  45. Gao, F. & Zhang, X. Mapping crop phenology in near real-time using satellite remote sensing: challenges and opportunities. J. Remote Sens. 2021, 8379391 (2021).
  46. Müller, M. Dynamic time warping. in Information Retrieval for Music and Motion. 69-84 (Springer, 2007).
  47. Bambach, N. et al. The Tree-crop Remote sensing of Evapotranspiration eXperiment (T-REX): a science-based path for sustainable water management and climate mitigation. Bull. Am. Meteorol. Soc. In the press (2023).
  48. Fisher, J. B. Hydrosat: towards daily, field-scale, global evapotranspiration from space. (2022).
  49. Polhamus, A., Fisher, J. B. & Tu, K. P. What controls the error structure in evapotranspiration models? Agric. For. Meteorol. 169, 12-24 (2013).
  50. Blankenau, P. A., Kilic, A. & Allen, R. An evaluation of gridded weather data sets for the purpose of estimating reference evapotranspiration in the United States. Agric. Water Manag. 242, 106376 (2020).
  51. Doherty, C. T. et al. Effects of meteorological and land surface modeling uncertainty on errors in winegrape ET calculated with SIMS. Irrig. Sci. 40, 515-530 (2022).
  52. Purdy, A., Fisher, J., Goulden, M. & Famiglietti, J. Ground heat flux: an analytical review of 6 models evaluated at 88 sites and globally. J. Geophys. Res. Biogeosci. 121, 3045-3059 (2016).
  53. Allen, R. G. et al. A recommendation on standardized surface resistance for hourly calculation of reference ETo by the FAO56 Penman-Monteith method. Agric. Water Manag. 81, 1-22 (2006).
  54. Jung, M. et al. The FLUXCOM ensemble of global landatmosphere energy fluxes. Sci. Data 6, 74 (2019).
  55. Reitz, M., Senay, G. B. & Sanford, W. E. Combining remote sensing and water-balance evapotranspiration estimates for the conterminous United States. Remote Sens. 9, 1181 (2017).
  56. Volk, J. et al. flux-data-qaqc: a Python package for energy balance closure and post-processing of eddy flux. Data. 6, 1-5 (2021).
  57. Evett, S. R. et al. The Bushland weighing lysimeters: a quarter century of crop ET investigations to advance sustainable irrigation. Trans. ASABE 59, 163-179 (2016).
  58. Abatzoglou, J. T. Development of gridded surface meteorological data for ecological applications and modelling. Int. J. Climatol. 33, 121-131 (2013).
  59. Kljun, N., Calanca, P., Rotach, M. W. & Schmid, H. P. A simple twodimensional parameterisation for Flux Footprint Prediction (FFP). Geosci. Model Dev. 8, 3695-3713 (2015).
  60. Xia, Y. et al. Continental-scale water and energy flux analysis and validation for the North American Land Data Assimilation System project phase 2 (NLDAS-2): 1. Intercomparison and application of model products. J. Geophys. Res. Atmos. 117, DO31O9 (2012).
  61. Foga, S. et al. Cloud detection algorithm comparison and validation for operational Landsat data products. Remote Sens. Environ. 194, 379-390 (2017).
  62. Rousseeuw, P. J. & Croux, C. Alternatives to the median absolute deviation. J. Am. Stat. Assoc. 88, 1273-1283 (1993).
  63. Harris, C. R. et al. Array programming with NumPy. Nature 585, 357-362 (2020).
  64. Seabold, S. & Perktold, J. Statsmodels: econometric and statistical modeling with Python. In Proc. 9th Python in Science Conference vol. 57 10-25080 (SciPy, 2010).
  65. Obrecht, N. A. Sample size weighting follows a curvilinear function. J. Exp. Psychol. Learn. Mem. Cogn. 45, 614 (2019).

شكر وتقدير

تم إنتاج بيانات OpenET المستخدمة في هذه الدراسة على Google Earth Engine ونعرب عن شكرنا لشركة Google، Inc. على الدعم والموارد الحاسوبية المستخدمة في إنتاج ومعالجة هذه البيانات. تم دعم العمل في هذا التحليل من قبل مؤسسة Walton Family Foundation؛ Lyda Hill Philanthropies؛ برنامج العلوم التطبيقية التابع لوكالة ناسا (NASA) (منحة NNX17AF53G، J.L.H؛ NNX12AD05A، F.M.); فريق علوم ناسا لاندسات التابع لهيئة المسح الجيولوجي الأمريكية (USGS) (رقم المنحة 140G0118C0007، J.L.H.); وحدات دراسات النظم البيئية التعاونية التابعة لهيئة المسح الجيولوجي الأمريكية (USGS) (منحة G23ACOO568، A.P.); معهد أبحاث موارد المياه التابع لهيئة المسح الجيولوجي الأمريكية (منحة G22AC00584-00، J.L.H.); مكتب تطبيقات المياه الغربية التابع لناسا (منحة 1669431، J.L.H؛ 80NSSC23K0836، C.P.); معهد أبحاث الزراعة التابع لجامعة ولاية كاليفورنيا (رقم المنحة 21-01-106، F.M.); المعهد الوطني للغذاء والزراعة McIntire Stennis (منحة MISZ-721160، Y.Y.); صندوق مكي للبحوث الصحراوية؛ ومحطة التجارب الزراعية في أيداهو ومحطة التجارب الزراعية في نبراسكا. يتم تقديم الدعم العيني من قبل الشركاء في مجتمعات الزراعة وإدارة المياه، وصندوق الدفاع البيئي وGoogle Earth Engine. نعترف ونشكر جهود جمع البيانات على المدى الطويل من قبل برنامج AmeriFlux، ووزارة الزراعة الأمريكية (USDA) خدمة أبحاث الزراعة، وهيئة المسح الجيولوجي الأمريكية (USGS)، وخدمة الإحصاءات الزراعية الوطنية التابعة لوزارة الزراعة الأمريكية (USDA). تشمل معرفات الكائنات الرقمية (DOIs) لمحطات AmeriFlux وUSGS المستخدمة في هذه الدراسة ما يلي: https://doi.org/10.17190/AMF/1436327، https://doi.org/10.17190/AMF/1436328، https://doi.org/10.17190/AMF/1418680، “https://doi.org/10.17190/AMF/1246137،https://doi.org/10.17190/AMF/1246025، https://doi. org/10.17190/AMF/1246026، https://doi.org/10.17190/AMF/1246027, https://doi.org/10.17190/AMF/1246028، https://doi.org/10.17190/AMF/1480317،https://doi.org/10.17190/AMF/1419513، https://doi. org/10.17190/AMF/1246040، https://doi.org/10.17190/AMF/1246031، “https://doi.org/10.17190/AMF/1246032، https://doi.org/10.17190/AMF/1246036،https://doi.org/10.17190/AMF/1246038، https://doi. org/10.17190/AMF/1246039، https://doi.org/10.17190/AMF/1246043، https://doi.org/10.17190/AMF/1660339، https://doi.org/10.17190/AMF/1246156،https://doi.org/10.17190/AMF/1246117، https://doi. org/10.17190/AMF/1419512، https://doi.org/10.17190/AMF/1246045، https://doi.org/10.17190/AMF/1246046، https://doi.org/10.17190/AMF/1246047،https://doi.org/10.17190/AMF/1246119، https://doi. org/10.17190/AMF/1246050، https://doi.org/10.17190/AMF/1246053، https://doi.org/10.17190/AMF/1246054، “https://doi.org/10.17190/AMF/1246051،https://doi.org/10.17190/AMF/1246052، https://doi. org/10.17190/AMF/1246056، https://doi.org/10.17190/AMF/1246057، https://doi.org/10.17190/AMF/1246058، https://doi.org/10.17190/AMF/1562389،https://doi.org/10.17190/AMF/1543379، https://doi.org/10.17190/AMF/1246065، https://doi.org/10.17190/AMF/1246066،https://doi.org/10.17190/AMF/1617696، “https://doi. org/10.17190/AMF/1498745، https://doi.org/10.17190/AMF/1634882،
https://doi.org/10.17190/AMF/1246070، https://doi.org/10.17190/AMF/1660346،https://doi.org/10.17190/AMF/1246074، https://doi. org/10.17190/AMF/1246076، https://doi.org/10.17190/AMF/1246079، https://doi.org/10.17190/AMF/1246128، https://doi.org/10.17190/AMF/1617715،https://doi.org/10.17190/AMF/1617716، “https://doi. org/10.17190/AMF/1246080، https://doi.org/10.17190/AMF/1246081، https://doi.org/10.17190/AMF/1246083، https://doi.org/10.17190/AMF/1419506،https://doi.org/10.17190/AMF/1480314، https://doi. org/10.17190/AMF/1246084، https://doi.org/10.17190/AMF/1246085، https://doi.org/10.17190/AMF/1246086، https://doi.org/10.17190/AMF/1246088،https://doi.org/10.17190/AMF/1246089، https://doi. org/10.17190/AMF/1246092، https://doi.org/10.17190/AMF/1418683، https://doi.org/10.17190/AMF/1246093، https://doi.org/10.17190/AMF/1419507،https://doi.org/10.17190/AMF/1419508، https://doi. org/10.17190/AMF/1419509، https://doi.org/10.17190/AMF/1617721، https://doi.org/10.17190/AMF/1617724، https://doi.org/10.17190/AMF/1375201،https://doi.org/10.17190/AMF/1419502، https://doi. org/10.17190/AMF/1419501، https://doi.org/10.17190/AMF/1419504، https://doi.org/10.17190/AMF/1246136، https://doi.org/10.17190/AMF/1246105،https://doi.org/10.17190/AMF/1246096، https://doi. org/10.17190/AMF/1418684، https://doi.org/10.17190/AMF/1246097، https://doi.org/10.17190/AMF/1246098، https://doi.org/10.17190/AMF/1246099،https://doi.org/10.17190/AMF/1246101، https://doi. org/10.17190/AMF/1246102، https://doi.org/10.17190/AMF/1246127، https://doi.org/10.17190/AMF/1246154، “https://doi.org/10.17190/AMF/1246104،https://doi.org/10.17190/AMF/1418685، https://doi. org/10.17190/AMF/1660351، https://doi.org/10.17190/AMF/1246148، https://doi.org/10.17190/AMF/1246149، https://doi.org/10.17190/AMF/1246140،https://doi.org/10.17190/AMF/1245984، https://doi. org/10.17190/AMF/1246109، https://doi.org/10.17190/AMF/1246111، https://doi.org/10.17190/AMF/1246112، https://doi.org/10.17190/AMF/1617728،https://doi.org/10.17190/AMF/1579721، https://doi. org/10.17190/AMF/1617732، https://doi.org/10.17190/AMF/1579723، https://doi.org/10.17190/AMF/1617735، https://doi.org/10.17190/AMF/1617737،https://doi.org/10.17190/AMF/1617741, https://doi. org/10.3133/sir20095079، https://doi.org/10.3133/sir20095079، https://doi.org/10.3133/sir20055288، https://doi.org/10.3133/سير20055288،https://doi.org/10.3133/sir20085116، https://doi. org/10.3133/sir20095079، https://doi.org/10.3133/sir20085116، https:// doi.org/10.5066/F7R49NZN، https://doi.org/10.5066/F7R49NZN، https://doi.org/10.5066/F79C6WM9, https://doi.org/10.5066/F79C6WM9https://doi.org/10.3133/pp1805، https://doi.org/10.5066/ P9NZ9XSP، https://doi.org/10.5066/P9NZ9XSP، https://doi. org/10.5066/P9NZ9XSP، https://doi.org/10.3133/sir20075078، https:// doi.org/10.3133/sir20075078، https://doi.org/10.3133/sir20085116، https://doi.org/10.3133/sir20075078 و https://doi.org/10.3133/السير 20075078. تم توفير التمويل لموارد بيانات أمري فلوكس من قبل مكتب العلوم بوزارة الطاقة الأمريكية. أي استخدام لأسماء التجارة أو الشركات أو المنتجات هو لأغراض وصفية فقط ولا يعني تأييدًا من الحكومة الأمريكية.

مساهمات المؤلفين

قام كل من F.S.M. و J.L.H. و J.M.V. و R.A. و M.A. و J.B.F. و A.K. و A.R. و G.B.S. و C.P. بتصميم وإدارة الدراسة؛ كتب J.M.V. و F.S.M. و M.A. و L.J. النص الرئيسي؛ أجرى J.M.V. التحليلات الإحصائية؛ أعد كل من C.M. و J.M.V. و B.M. و T.O. و C.D. و T.W. البيانات المقاسة أو بيانات إدخال النموذج، أو قاموا بتشغيل النماذج؛ طور كل من F.S.M. و R.A. و M.A. و J.B.F. و A.K. و A.R. و G.B.S. و J.L.H. و C.M. و W.C. و C.T.D. و M.F. و A.G. و C.H. و G.H. و L.J. و Y.K. و K.K. و S.O.-S. و G.E.L.P. و A.P. و P.R. و Y.Y. و L.L. و B.C.d.A. النماذج وبنية OpenET؛ راجع كل من J.M.V. و M.A. و F.S.M. و L.J. و R.A. و J.B.F. و J.L.H. و A.K. و G.B.S. و T.O. و B.M. و A.R. و M.F. و T.W. النصوص والأشكال.

المصالح المتنافسة

يعلن المؤلفون عدم وجود مصالح متنافسة.

معلومات إضافية

البيانات الموسعة متاحة لهذا البحث في
https://doi.org/10.1038/s44221-023-00181-7.
معلومات إضافية النسخة الإلكترونية تحتوي على مواد إضافية متاحة فيhttps://doi.org/10.1038/s44221-023-00181-7.
يجب توجيه المراسلات والطلبات للحصول على المواد إلى جون م. فولت.
تُعرب مجلة Nature Water عن شكرها لتيلدن مايرز، دينيس بالدوكي، والمراجعين الآخرين المجهولين، على مساهمتهم في مراجعة هذا العمل.
معلومات إعادة الطباعة والتصاريح متاحة علىwww.nature.com/reprints.
ملاحظة الناشر: تظل شركة سبرينجر ناتشر محايدة فيما يتعلق بالمطالبات القضائية في الخرائط المنشورة والانتماءات المؤسسية.
الوصول المفتوح هذه المقالة مرخصة بموجب رخصة المشاع الإبداعي النسب 4.0 الدولية، التي تسمح بالاستخدام والمشاركة والتكيف والتوزيع وإعادة الإنتاج بأي وسيلة أو صيغة، طالما أنك تعطي الائتمان المناسب للمؤلفين الأصليين والمصدر، وتوفر رابطًا لرخصة المشاع الإبداعي، وتوضح ما إذا تم إجراء تغييرات. الصور أو المواد الأخرى من طرف ثالث في هذه المقالة مشمولة في رخصة المشاع الإبداعي الخاصة بالمقالة، ما لم يُشار إلى خلاف ذلك في سطر الائتمان للمواد. إذا لم تكن المادة مشمولة في رخصة المشاع الإبداعي الخاصة بالمقالة وكان استخدامك المقصود غير مسموح به بموجب اللوائح القانونية أو يتجاوز الاستخدام المسموح به، فستحتاج إلى الحصول على إذن مباشرة من صاحب حقوق الطبع والنشر. لعرض نسخة من هذه الرخصة، قم بزيارةhttp://creativecommons.org/licenses/by/4.0/.
هذا عمل حكومي أمريكي وليس محميًا بحقوق الطبع والنشر في الولايات المتحدة؛ قد تنطبق حماية حقوق الطبع والنشر الأجنبية 2024
معهد أبحاث الصحراء، رينو، نيفادا، الولايات المتحدة الأمريكية. مركز أبحاث ناسا أيمس، موفيت فيلد، كاليفورنيا، الولايات المتحدة الأمريكية. جامعة ولاية كاليفورنيا مونتيري باي، سيسايد، كاليفورنيا، الولايات المتحدة الأمريكية. جامعة أيداهو، كيمبرلي، أيداهو، الولايات المتحدة الأمريكية. خدمة البحوث الزراعية التابعة لوزارة الزراعة الأمريكية، بيلتسفيل، ماريلاند، الولايات المتحدة الأمريكية. جامعة كاليفورنيا، لوس أنجلوس، لوس أنجلوس، كاليفورنيا، الولايات المتحدة الأمريكية. جامعة نبراسكا – لينكولن، لينكولن، نبراسكا، الولايات المتحدة الأمريكية. الجامعة الفيدرالية في ريو غراندي دو سول، بورتو أليغري، RS، البرازيل. مركز مراقبة موارد الأرض والعلوم التابع لمسح الجيولوجيا الأمريكي، مركز علوم التكيف المناخي في شمال وسط الولايات المتحدة، فورت كولينز، كولورادو، الولايات المتحدة الأمريكية. شركة KBR، بموجب عقد مع المسح الجيولوجي الأمريكي، مركز مراقبة وعلوم موارد الأرض، سيوكس فولز، داكوتا الجنوبية، الولايات المتحدة الأمريكية. مركز مارشال لرحلات الفضاء التابع لناسا، هنتسفيل، ألاباما، الولايات المتحدة الأمريكية. مختبر الدفع النفاث التابع لناسا، باسادينا، كاليفورنيا، الولايات المتحدة الأمريكية. جامعة كاليفورنيا في بيركلي، بيركلي، الولايات المتحدة الأمريكية. خدمة البحوث الزراعية التابعة لوزارة الزراعة الأمريكية، ديفيس، كاليفورنيا، الولايات المتحدة الأمريكية. إنوفيت!، شركة بموجب عقد مع المسح الجيولوجي الأمريكي مركز مراقبة وعلوم موارد الأرض، سيوكس فولز، داكوتا الجنوبية، الولايات المتحدة الأمريكية. جامعة ولاية ميسيسيبي، ستاركفيل، ميسيسيبي، الولايات المتحدة الأمريكية. البريد الإلكتروني: john.volk@dri.edu

الغابات دائمة الخضرة

الشكل البياني الموسع 1 | المناخ الشهري للبيانات المقارنة بين ET النموذجي والملاحظ لمواقع الغابات دائمة الخضرة. يوضح الرسم الفرعي (أ) المناخ الشهري للبيانات المقارنة بين OpenET وبرج التدفق من المواقع الحرجية الدائمة الخضرة. يظهر الرسم الفرعي (ب) المتبقي من متوسط ET الشهري (النموذج ناقص متوسط تدفق ET المغلق). غير مغلق
تشير التسميات المغلقة وغير المغلقة إلى تدفق برج ET قبل وبعد تصحيح توازن الطاقة. تمثل الخطوط المتقطعة متوسط تدفق ET المغلق زائد اثنين من الأخطاء المعيارية للمتوسط ومتوسط تدفق ET غير المغلق ناقص اثنين من الأخطاء المعيارية للمتوسط.

الغابات المختلطة

الشكل البياني الموسع 2 | المناخ الشهري للبيانات المودلة والملاحظة لمواقع الغابات المختلطة. يوضح الرسم الفرعي (أ) المناخ الشهري للبيانات المودلة OpenET. وبرج التدفق من مواقع الغابات المختلطة. يُظهر الجزء الفرعي (ب) المتبقي من متوسط ET الشهري (النموذج ناقص متوسط تدفق ET المغلق). غير مغلق
تشير التسميات المغلقة وغير المغلقة إلى تدفق برج ET قبل وبعد تصحيح توازن الطاقة. تمثل الخطوط المتقطعة متوسط تدفق ET المغلق زائد اثنين من الأخطاء المعيارية للمتوسط ومتوسط تدفق ET غير المغلق ناقص اثنين من الأخطاء المعيارية للمتوسط.

المراعي

الشكل 3 من البيانات الموسعة | المناخ الشهري للنماذج المقترنة والملاحظات
ET لمواقع المراعي. توضح الفقرة (أ) المناخ الشهري لبيانات OpenET المزدوجة وبرج التدفق ET من مواقع المراعي. يوضح الرسم الفرعي (ب) المتبقي من متوسط ET الشهري (النموذج ناقص متوسط تدفق ET المغلق). غير مغلق
تشير التسميات المغلقة وغير المغلقة إلى تدفق برج ET قبل وبعد تصحيح توازن الطاقة. تمثل الخطوط المتقطعة متوسط تدفق ET المغلق زائد اثنين من الأخطاء المعيارية للمتوسط ومتوسط تدفق ET غير المغلق ناقص اثنين من الأخطاء المعيارية للمتوسط.

الأراضي الشجرية

الشكل 4 من البيانات الموسعة | المناخ الشهري للنماذج المقترنة والملاحظات
ET لمواقع الشجيرات. يوضح الرسم الفرعي (أ) المناخ الشهري لبيانات OpenET المزدوجة وبرج التدفق من مواقع الشجيرات. يوضح الرسم الفرعي (ب) المتبقي من متوسط ET الشهري (النموذج ناقص متوسط تدفق ET المغلق). غير مغلق
تشير التسميات المغلقة وغير المغلقة إلى تدفق برج ET قبل وبعد تصحيح توازن الطاقة. تمثل الخطوط المتقطعة متوسط تدفق ET المغلق زائد اثنين من الأخطاء المعيارية للمتوسط ومتوسط تدفق ET غير المغلق ناقص اثنين من الأخطاء المعيارية للمتوسط.

الأراضي الرطبة/المناطق الساحلية

مجموعة
ديزاكسي eeMETRIC
جي سيبال
PT-JPL
SSEBop
مغلق + 2SE
مغلق
غير مغلق
– – غير مغلق – 2SE

شهر

الشكل البياني الممتد 5 | المناخ الشهري للبيانات المودلة والملاحظة لمعدل التبخر والنتح في المواقع الرطبة والمناطق النهرية. يوضح الرسم الفرعي (أ) المناخ الشهري للبيانات المودلة المقترنة بـ OpenET وبرج التدفق من المواقع الرطبة والمناطق النهرية. يُظهر الجزء الفرعي (ب) المتبقي من متوسط ET الشهري (النموذج ناقص متوسط تدفق ET المغلق).
تشير التسميات غير المغلقة والمغلقة إلى تدفق برج ET قبل وبعد تصحيح توازن الطاقة. تمثل الخطوط المتقطعة متوسط تدفق ET المغلق زائد اثنين من الأخطاء المعيارية للمتوسط ومتوسط تدفق ET غير المغلق ناقص اثنين من الأخطاء المعيارية للمتوسط.
الشكل 6 من البيانات الموسعة | المناخ الشهري لنموذج ET باستخدام جميع بكسلات الأراضي الزراعية. المناخ الشهري لـ OpenET أعضاء المجموعة والمتوسط الجماعي باستخدام جميع بيانات ET الشهرية لجميع البكسلات التي تم تصنيفها كأراضٍ زراعية لكل عام من 2016 إلى 2022.
متوسط عدد النماذج في التجميع
الشكل البياني الممتد 7| انظر الصفحة التالية للتعليق.
الشكل البياني الموسع 7| التحليل المكاني لحدوث القيم الشاذة في مجموعة نماذج بكسلات الأراضي الزراعية. يظهر الرسم الفرعي (أ) الفروق المكانية بين متوسط ​​فترة النمو لمجموعة OpenET ET لفترة النمو (من أبريل إلى أكتوبر) لبكسلات الأراضي الزراعية باستخدام طريقة إزالة القيم الشاذة بالانحراف المطلق الوسيط (MAD) والمتوسط الحسابي البسيط (SAM)؛ تم استخدام ET الشهري من 2016-2022
لبناء الخريطة. يظهر الرسم الفرعي (ب) متوسط ​​عدد النماذج المستخدمة في المجموعة بعد إزالة القيم الشاذة باستخدام جميع بيانات الأشهر لفترة النمو لبكسلات الأراضي الزراعية. تشير القيمة ستة إلى أنه لم يتم تحديد أي نموذج كقيمة شاذة، بينما أربعة هي الحد الأدنى حيث تم إزالة نموذجين كحد أقصى كقيم شاذة قبل أخذ المتوسط الجماعي.
الشكل البياني الموسع الفرق المكاني بين متوسط ​​فترة النمو ET لكل نموذج من القيمة الجماعية في بكسلات الأراضي الزراعية. الفرق بين متوسط ​​فترة النمو (من أبريل إلى أكتوبر) ET من كل نموذج OpenET
ناقص المتوسط الجماعي باستخدام جميع بيانات الأشهر من جميع البكسلات التي تم تصنيفها كأراضي زراعية لكل عام من 2016-2022. انظر المناقشة التكميلية 4 لمناقشة خطوط Landsat المعروضة بواسطة geeSEBAL.

natureportfolio

المؤلف(ون) المعني(ون):
جون فوك
آخر تحديث من قبل المؤلف(ين): 27 نوفمبر 2023

ملخص التقرير

تسعى Nature Portfolio لتحسين إمكانية إعادة إنتاج العمل الذي ننشره. يوفر هذا النموذج هيكلًا للاتساق والشفافية في التقرير. لمزيد من المعلومات حول سياسات Nature Portfolio، انظر سياسات التحرير وقائمة مراجعة سياسة التحرير.

الإحصائيات

لجميع التحليلات الإحصائية، تأكد من أن العناصر التالية موجودة في أسطورة الشكل، أسطورة الجدول، النص الرئيسي، أو قسم الطرق.

تم التأكيد

حجم العينة الدقيقة ( ) لكل مجموعة/شرط تجريبي، معطاة كرقم منفصل ووحدة قياس
بيان حول ما إذا كانت القياسات مأخوذة من عينات متميزة أو ما إذا كانت نفس العينة تم قياسها عدة مرات
اختبار(ات) إحصائية المستخدمة وما إذا كانت أحادية أو ثنائية الجانب
يجب وصف الاختبارات الشائعة فقط بالاسم؛ وصف تقنيات أكثر تعقيدًا في قسم الطرق.
وصف لجميع المتغيرات المرافقة التي تم اختبارها
وصف لأي افتراضات أو تصحيحات، مثل اختبارات الطبيعية والتعديل لمقارنات متعددة
وصف كامل للمعلمات الإحصائية بما في ذلك الاتجاه المركزي (مثل المتوسطات) أو تقديرات أساسية أخرى (مثل معامل الانحدار) والتباين (مثل الانحراف المعياري) أو تقديرات عدم اليقين المرتبطة (مثل فترات الثقة)
لاختبار فرضية العدم، إحصائية الاختبار (مثل F، ) مع فترات الثقة، أحجام التأثير، درجات الحرية و القيمة المذكورة أعطِ القيم كقيم دقيقة كلما كان ذلك مناسبًا.
لتحليل بايزي، معلومات حول اختيار الأوليات وإعدادات سلسلة ماركوف مونت كارلو
للتصاميم الهرمية والمعقدة، تحديد المستوى المناسب للاختبارات والتقارير الكاملة للنتائج
تقديرات أحجام التأثير (مثل حجم كوهين، ، حجم بيرسون، )، مشيرًا إلى كيفية حسابها
تحتوي مجموعتنا على الإنترنت حول الإحصائيات لعلماء الأحياء على مقالات حول العديد من النقاط أعلاه.

البرمجيات والرموز

معلومات السياسة حول توفر كود الكمبيوتر

جمع البيانات
تم معالجة بيانات التبخر والنتح المقاسة في الموقع التي تم تحليلها خلال الدراسة الحالية باستخدام حزمة بايثون “flux-data-qaqc”، الإصدار 0.1.6 (https://github.com/Open-ET/flux-data-qaqc).
تحليل البيانات
تم استخدام حزمة بايثون “flux-data-footprint” لإنشاء بصمات تدفق ديناميكية زمنية لعينات بيانات ET اليومية والشهرية (https://github.com/Open-ET/flux-data-footprint). تم استخدام حزم بايثون Numpy (الإصدار 1.17.2) وstatsmodels (الإصدار 0.12.1) لتحليل البيانات خلال الدراسة الحالية.
بالنسبة للمخطوطات التي تستخدم خوارزميات أو برمجيات مخصصة تكون مركزية للبحث ولكن لم يتم وصفها بعد في الأدبيات المنشورة، يجب أن تكون البرمجيات متاحة للمحررين والمراجعين. نشجع بشدة على إيداع الكود في مستودع مجتمعي (مثل GitHub). انظر إرشادات Nature Portfolio لتقديم الكود والبرمجيات لمزيد من المعلومات.

البيانات

معلومات السياسة حول توفر البيانات
يجب أن تتضمن جميع المخطوطات بيانًا حول توفر البيانات. يجب أن يوفر هذا البيان المعلومات التالية، حيثما ينطبق:
  • رموز الوصول، معرفات فريدة، أو روابط ويب لمجموعات البيانات المتاحة للجمهور
  • وصف لأي قيود على توفر البيانات
  • بالنسبة لمجموعات البيانات السريرية أو بيانات الطرف الثالث، يرجى التأكد من أن البيان يتماشى مع سياستنا
zenodo.7636781. بيانات ET لنموذج OpenET التي تم تحليلها خلال الدراسة الحالية متاحة في مستودع Zenodo، مع المعرف http://dx.doi.org/10.5281/ zenodo. 10119477.

المشاركون في البحث البشري

معلومات السياسة حول الدراسات التي تشمل المشاركين في البحث البشري والجنس والنوع في البحث.
التقرير عن الجنس والنوع لم تتضمن الدراسة الحالية مشاركين بشريين، بياناتهم، أو موادهم البيولوجية.
خصائص السكان لم تتضمن الدراسة الحالية مشاركين بشريين، بياناتهم، أو موادهم البيولوجية.
التجنيد لم تتضمن الدراسة الحالية مشاركين بشريين، بياناتهم، أو موادهم البيولوجية.
الإشراف الأخلاقي لم تتضمن الدراسة الحالية مشاركين بشريين، بياناتهم، أو موادهم البيولوجية.
لاحظ أنه يجب أيضًا تقديم معلومات كاملة حول الموافقة على بروتوكول الدراسة في المخطوطة.

التقارير الخاصة بالمجال

يرجى اختيار الخيار أدناه الذي يناسب بحثك بشكل أفضل. إذا لم تكن متأكدًا، اقرأ الأقسام المناسبة قبل اتخاذ قرارك.
علوم الحياة العلوم السلوكية والاجتماعية
العلوم البيئية والتطورية والبيئية
لنسخة مرجعية من الوثيقة مع جميع الأقسام، انظر nature.com/documents/nr-reporting-summary-flat.pdf

تصميم دراسة العلوم البيئية والتطورية والبيئية

يجب على جميع الدراسات الإفصاح عن هذه النقاط حتى عندما يكون الإفصاح سلبيًا.
وصف الدراسة تركز الدراسة الحالية على المقارنات واحد لواحد بين بيانات التبخر والنتح كما تم نمذجتها بواسطة طرق الاستشعار عن بعد وكما تم قياسها على الأرض؛ تم استخدام مقاييس جودة الملاءمة المعروفة لتقييم بيانات النموذج مقابل البيانات المقاسة. تم ربط البيانات النمذجة والمقاسة بناءً على سجلات متداخلة زمنياً في محطات قياس متعددة. تم تجميع نتائج مقاييس الدقة حسب خصائص مشتركة مختلفة للمحطات، باستخدام المتوسط المرجح. تم رسم بيانات النموذج طويلة الأجل للموديلات الفردية وتم طرحها من المتوسط الجماعي للنموذج.
عينة البحث جاءت بيانات التبخر والنتح اليومية والشهرية المقاسة في الدراسة الحالية من مجموعة بيانات عامة متاحة هنا: http://dx.doi.org/10.5281/zenodo.7636781. تم جمع القياسات المستخدمة بين 1995-2021. تم إنشاء بيانات نموذج OpenET وتم ربطها بالقياسات اليومية والشهرية، ومع ذلك لم تكن بيانات النموذج متاحة قبل عام 2001. كان العدد الإجمالي للمحطات التي تحتوي على بيانات مرتبطة 152، ومن بينها كان هناك 16,444 يومًا و4,107 أشهر من البيانات المرتبطة.
استراتيجية العينة كإجراء احترازي، طلبنا حدًا أدنى من 3 أشهر من البيانات المرتبطة لكل محطة قياس ليتم تضمينها في مقاييس دقة المتوسط المرجح. لتجنب تحريف مقاييس المتوسطات المجمعة، قمنا بوزن كل محطة بجذر عدد البيانات المرتبطة.
جمع البيانات تمت معالجة بيانات ET المقاسة مسبقًا وأرشفتها علنًا على Zenodo (http://dx.doi.org/10.5281/zenodo.7636781). تم إنشاء بيانات ET للنموذج للدراسة باستخدام الإصدار الحالي من OpenET.
التوقيت والنطاق المكاني تم جمع البيانات المقاسة بشكل أساسي من أنظمة تباين الإدي وجرى معالجتها لاحقًا إلى فترات مجمعة يومية وشهرية. تم أخذ عينات من بيانات ET للنموذج على فترات يومية (تاريخ مرور القمر الصناعي) وشهرية في محطات القياس باستخدام بصمات بكسل تم تحديدها إما من اتجاه الرياح طويل الأجل أو من نموذج توقع بصمة التدفق القائم على الفيزياء. نادرًا ما تجاوزت بصمات التدفق حجم شبكة (دقة 30 م). توجد فجوات في البيانات المقاسة بسبب توقف تشغيل المستشعر أو بيانات خاطئة، وتوجد فجوات في بيانات النموذج بسبب تغطية السحب.
استبعاد البيانات لا توجد بيانات متاحة لنا تم استبعادها من التحليلات خلال الدراسة الحالية.
إعادة الإنتاجية تم جعل خطوات معالجة البيانات لكل من البيانات المودلة والمقاسة المستخدمة في الدراسة الحالية قابلة لإعادة الإنتاج من خلال استخدامنا لبرامج مفتوحة المصدر موثقة جيدًا. قمنا بتطوير كود بايثون لمعالجة بيانات التبادل الحراري وتطوير البصمة. وبالمثل، تم توليد بيانات OpenET المودلة باستخدام النماذج التشغيلية التي هي مفتوحة المصدر وبياناتها متاحة أيضًا للجمهور من خلال آليات مختلفة بما في ذلك واجهة برمجة تطبيقات OpenET وكاتالوج بيانات Google Earth Engine. كانت الطرق الإحصائية المستخدمة في الدراسة محدودة بتقنيات بسيطة، وتم اختبار النتائج بشكل مستقل من قبل عدة أعضاء من مجموعة OpenET باستخدام حزم إحصائية مختلفة مثل بايثون ومايكروسوفت إكسل.
العشوائية لم يكن من الممكن تطبيق عشوائية البيانات إلى مجموعات حيث تم تعريف المجموعات بواسطة الخصائص البيوفيزيائية مثل المناخ ونوع تغطية الأرض.
التعمية لم تستخدم هذه الدراسة تعمية البيانات بسبب الكمية المحدودة من بيانات التبخر والنتح المقاسة عالية الجودة المتاحة لنا. ومع ذلك، تم الاحتفاظ ببيانات إضافية لم تتم مقارنتها بعد مع نماذج OpenET خارج هذه الدراسة من أجل مقارنة عمياء مستقبلية وتقييم دقة OpenET.

التقارير عن مواد وأنظمة وطرق محددة

نحتاج إلى معلومات من المؤلفين حول بعض أنواع المواد والأنظمة التجريبية والطرق المستخدمة في العديد من الدراسات. هنا، حدد ما إذا كانت كل مادة أو نظام أو طريقة مدرجة ذات صلة بدراستك. إذا لم تكن متأكدًا مما إذا كان عنصر القائمة ينطبق على بحثك، اقرأ القسم المناسب قبل اختيار رد.
المواد والأنظمة التجريبية الطرق
غير متاح المشاركة في الدراسة غير متاح المشاركة في الدراسة
الأجسام المضادة
ChIP-seq
علم الحفريات وعلم الآثار

  1. بيانات التبخر والنتح المقاسة في الموقع التي تم تحليلها خلال الدراسة الحالية متاحة في مستودع Zenodo، مع المعرف http://dx.doi.org/10.5281/

Journal: Nature Water, Volume: 2, Issue: 2
DOI: https://doi.org/10.1038/s44221-023-00181-7
Publication Date: 2024-01-15

Assessing the accuracy of OpenET satellite-based evapotranspiration data to support water resource and land management applications

Received: 21 June 2023
Accepted: 30 November 2023
Published online: 15 January 2024

Check for updates

Abstract

John M. Volk © , Justin L. Huntington , Forrest S. Melton , Richard Allen , Martha Anderson , Joshua B. Fisher © , Ayse Kilic , Anderson Ruhoff © , Gabriel B. Senay , Blake Minor , Charles Morton , Thomas Ott , Lee Johnson (1) , Bruno Comini de Andrade , Will Carrara , Conor T. Doherty , Christian Dunkerly ( ) , MacKenzie Friedrichs ( , Alberto Guzman , Christopher Hain , Gregory Halverson , Yanghui Kang (B , Kyle Knipper (B , Leonardo Laipelt , Samuel Ortega-Salazar , Christopher Pearson , Gabriel E. L. Parrish , Adam Purdy , Peter ReVelle , Tianxin Wang & Yun Yang

Abstract

Remotely sensed evapotranspiration (ET) data offer strong potential to support data-driven approaches for sustainable water management. However, practitioners require robust and rigorous accuracy assessments of such data. The OpenET system, which includes an ensemble of six remote sensing models, was developed to increase access to field-scale ( 30 m ) ET data for the contiguous United States. Here we compare OpenET outputs against data from 152 in situ stations, primarily eddy covariance flux towers, deployed across the contiguous United States. Mean absolute error at cropland sites for the OpenET ensemble value is 15.8 mm per month ( of mean observed ET), mean bias error is -5.3 mm per month ( ) and is 0.9 . Results for shrublands and forested sites show higher inter-model variability and lower accuracy relative to croplands. High accuracy and multi-model convergence across croplands demonstrate the utility of a model ensemble approach, and enhance confidence among ET data practitioners, including the agricultural water resource management community.

Accurate evapotranspiration (ET) data are essential for assessing the surface energy and water balance, the carbon cycle and the management of water resources . ET is the sum of the flux of water vapour from soil (evaporation) and through vegetation (transpiration) to the atmosphere. ET constitutes the second largest component of the terrestrial water balance, after precipitation. The usefulness of spatially contiguous mapping of ET, particularly over irrigated agricultural lands, has been amplified by drought, climate change, and high rates of human water withdrawal and agricultural consumption, leaving
many aquifers and water reservoirs in the western United States at all-time-low levels . Satellite-based remote sensing of ET (RSET) offers a powerful approach for mapping ET over large geographic regions at semi-continuous timescales . Until recently, the availability of RSET data at spatial scales relevant for water resources management has been limited by cost and computational requirements.
OpenET employs six state-of-the-art satellite based RSET models, that is, ALEXI/DisALEXI , eeMETRIC , geeSEBAL , PT-JPL , SIMS and SSEBop , that have been widely applied and evaluated in the United
Fig.1 | Map of in situ ET measurement sites. Map of the locations of in situ ET stations used to evaluate OpenET, including their general land cover type and Köppen-Geiger (KG) climate zones . White areas represent climate zones that did not contain any cropland sites and were excluded from the analysis.
Climate zone abbreviations are defined as follows: cold and hot semi-arid steppe (Bsk + Bsh); hot and cold desert (Bwh + Bwk); humid subtropical (Cfa); hot- and warm-summer Mediterranean (Csa + Csb); and hot- and warm-summer humid continental (Dfa + Dfb).
States for a range of water management and agricultural applications. The models are applied on the Google Earth Engine cloud-based platform to provide historical and near real-time ET data at subfield scales ( pixels) over the western United States . Five of the RSET models constrain components of the surface energy balance (SEB) using land surface temperature (LST) primarily derived from Landsat Collection 2, along with gridded weather data, and land cover datasets. The sixth model, SIMS, assumes well-watered conditions and computes crop coefficients based on vegetation density, derived from satellite surface reflectance values, along with a gridded soil water balance model. The models composing OpenET have been used by water managers, farmers and governmental organizations for irrigation scheduling, water accounting and allocation, and water rights administration . The OpenET platform provides an unprecedented level of accessibility to RSET data through its public online data explorer interface-including querying satellite ET within individually vectorized field boundaries. All six RSET models in OpenET operate automatically, including any required calibrations, which permits rapid calculations for the more than 100,000 Landsat images processed so far across the 23 western-most states in the contiguous United States. As the number of applications of RSET data for sustainable land and water resources management grow, it is important for practitioners to have information on the accuracy of RSET data across land cover types, climatic zones and agricultural production practices .
In this Analysis, we present a large-scale benchmark assessment of the accuracy of OpenET data using a well-curated publicly archived dataset of in situ ET measurements from 152 stations (141 eddy covariance (EC) systems, 7 Bowen ratio systems and 4 lysimeters), over a variety of regions, climates and land cover types , collectively comprising years of paired model-measurement ET data (Fig. 1). The EC technique is generally viewed as the best available method for continuous measurement of in situ energy and heat flux at spatial scales that approach satellite-based retrievals , although we acknowledge the associated data uncertainties and made efforts to reduce them . In addition to evaluation of individual model accuracies, we evaluated the OpenET ensemble ET value, computed as the mean of all models after flagging and removal of up to two outliers using the median absolute
deviation (MAD) approach . The generation of an ensemble value is a widely used technique to combine outputs from diverse models, each having their own behaviour and random error . It also facilitates applications such as irrigation scheduling and water rights administration, where practitioners require a single value for use in management of water resources . The publicly archived in situ flux dataset allows for reproducibility and benchmarking of future OpenET model versions or other RSET data.
ET data computed from micrometeorological measurements at EC sites were obtained from a variety of sources, primarily AmeriFlux . Supplementary Table 1 provides a full list of stations used in the study including land cover type, site principal investigators, Digital Object Identifiers (DOIs) and other metadata. Flux data were carefully postprocessed, including gap-filling, screening for energy balance closure error and data completeness, and visual data quality assessments. Flux data that passed quality control and showed limited energy balance closure error were included in the study and underwent closure correction following the FLUXNET2015/ONEFlux approach for daily averaged fluxes . We refer to EC data as ‘ECET’ throughout the article. Closed ECET data were considered to be most representative of actual . To sample RSET pixels for comparison with ECET, flux footprints were developed for each station. Flux footprints are twodimensional mappings of the areal extent of a station’s source area, that is, the area on the ground that contributes to fluxes measured by the tower instrumentation. Refer to Methods and Volk et al. for details on flux data processing and footprint mapping methods used. Additional discussion of uncertainty in EC data and steps taken to limit that uncertainty are provided in Supplementary Discussion 1. An overview of the satellite-driven ET models in the OpenET ensemble is provided in Methods.
The discussion of statistical results that follows focuses on comparisons between monthly aggregated ECET and RSET. Although accuracy assessments were conducted using daily (date of overpass) data and monthly total ET aggregated to growing season and annual periods, our discussion focuses on monthly results for several reasons: monthly ET has utility for longer-term water accounting and planning; uncertainties in EC data due to closure and other factors are reduced
Table 1 | Smmary statistics between modelled and observed monthly ET for cropland sites
Land cover type Statistic Ensemble DisALEXI eeMETRIC geeSEBAL PT-JPL SIMS SSEBop sites data points
All crops, mean station ET of 91 (mm per month) Slope 0.92 0.92 0.95 0.85 0.91 0.99 0.95 53 1,652
MBE (mm) -5.27 (-5.8%) -7.72 (-8.4%) -2.44 (-2.7%) -12.18 (-13.3%) -2.9 (-3.2%) 4.32 (4.7%) -6.08 (-6.7%) 44 1,638
MAE (mm) 15.84 (17.3%) 19.91 (21.8%) 21.23 (23.2%) 22.69 (24.8%) 18.12 (19.8%) 17.93 (19.6%) 22.4 (24.5%) 44 1,638
RMSE (mm) 20.44 (22.4%) 25.35 (27.7%) 26.97 (29.5%) 29.05 (31.8%) 23.67 (25.9%) 23.1 (25.3%) 27.72 (30.3%) 44 1,638
0.9 0.86 0.83 0.83 0.87 0.86 0.85 53 1,652
Annual crops, mean station ET of 85 (mm per month) Slope 0.93 0.92 0.98 0.85 0.9 1.01 0.92 42 1,446
MBE (mm) -5.11 (-6.0%) -8.18 (-9.6%) 0.23 (0.3%) -12.38 (-14.6%) -3.77 (-4.4%) 6.27 (7.4%) -9.13 (-10.7%) 36 1,436
MAE (mm) 15.26 (17.9%) 20.09 (23.6%) 20.44 (24.0%) 22.52 (26.5%) 17.0 (20.0%) 17.48 (20.5%) 21.93 (25.8%) 36 1,436
RMSE (mm) 19.71 (23.2%) 25.68 (30.2%) 26.17 (30.8%) 28.67 (33.7%) 22.31 (26.2%) 22.49 (26.4%) 27.14 (31.9%) 36 1,436
0.9 0.84 0.83 0.82 0.87 0.85 0.84 42 1,446
Orchards, mean station ET of 126 (mm per month) Slope 0.87 0.88 0.81 0.84 0.88 0.93 0.97 5 141
MBE (mm) -11.9 (-9.4%) -11.02 (-8.7%) -20.66 (-16.4%) -15.11 (-12.0%) -7.39 (-5.8%) -3.47 (-2.7%) -3.69 (-2.9%) 5 141
MAE (mm) 21.18 (16.8%) 22.2 (17.6%) 28.19 (22.3%) 24.9 (19.7%) 24.43 (19.3%) 22.95 (18.2%) 20.18 (16.0%) 5 141
RMSE (mm) 27.89 (22.1%) 27.86 (22.1%) 35.26 (27.9%) 32.77 (25.9%) 31.67 (25.1%) 30.49 (24.1%) 27.26 (21.6%) 5 141
0.91 0.89 0.89 0.88 0.88 0.86 0.89 5 141
Vineyards, mean station ET of 112 (mm per month) Slope 1.02 1.02 0.95 0.95 1.09 0.92 1.25 3 61
MBE (mm) 5.27 (4.7%) 4.99 (4.5%) -4.62 (-4.1%) -3.76 (-3.4%) 17.95 (16.0%) -7.71 (-6.9%) 31.88 (28.5%) 3 61
MAE (mm) 13.66 (12.2%) 13.01 (11.6%) 18.73 (16.7%) 20.72 (18.5%) 21.53 (19.2%) 14.43 (12.9%) 33.22 (29.7%) 3 61
RMSE (mm) 16.23 (14.5%) 15.87 (14.2%) 22.1 (19.7%) 27.2 (24.3%) 27.19 (24.3%) 17.34 (15.5%) 36.56 (32.6%) 3 61
0.9 0.9 0.81 0.73 0.83 0.88 0.84 3 61

of the weighted mean closed flux tower ET. Note, there were three additional vegetable crop sites included in the combined crop group, which alone did not meet our data requirements for statistical analyses .
at the monthly (compared with daily) timescale, and OpenET directly provides daily and monthly ET, along with data services that allow users to compute ET at other aggregation periods. Accuracy results are provided for daily, monthly, seasonal and annual timesteps in Supplementary Tables 2-6, and accuracy metrics for daily timesteps should be consulted for applications of ET data at timesteps of 1-15 days. Five well-known statistical metrics were used to evaluate OpenET accuracy (for equations, see Methods): the linear regression slope forced through the origin which measures bias (Slope), mean bias error (MBE), mean absolute error (MAE), root-mean-square error (RMSE) and the coefficient of determination ( ). Regression results with a non-zero intercept for monthly data are provided in Supplementary Table 7.

Performance over all agricultural flux sites

Of all the general land cover types sampled, OpenET models showed the strongest agreement with ECET collected in agricultural settings. For 44 agricultural sites combined, eeMETRIC, SIMS and PT-JPL showed the least bias in terms of MBE, all less than -4.5 mm per month or of the mean ECET (Table 1). The ensemble value had a slightly higher magnitude bias of -5.3 mm per month, or of the mean ECET. The ensemble value outperformed each individual model in terms of MAE 15.9 mm per month ( of the mean ECET), RMSE 20.4 mm per month ( ) and ( 0.90 ). In comparison, MAE from individual models ranged from 17.9 to 22.7 mm , RMSE from 23.1 to 29.1 mm per month and from 0.83 to 0.87 with smallest errors from PT-JPL, SIMS and DisALEXI.
ET data from the individual RSET models were generally linearly related to ECET, with PT-JPL and SIMS exhibiting some curvature due to seasonally varying biases (Fig. 2). Many of the models underestimated ET during the cold season relative to the ECET, leading to the slightly low bias in the ensemble ET value (Table 2). To investigate seasonal variability in model accuracy, we pooled all monthly paired
(model-measured) ET to generate monthly climatologies for major land cover classifications (Fig. 3 and Extended Data Figs. 1-5). The range between unclosed and closed ECET provides one measure of the uncertainty in the in situ data .
For most months, the multi-model ensemble ET value was well bounded between the closed and unclosed mean ECET for cropland sites, while individual ensemble members showed more seasonal bias. In spring, SSEBop and eeMETRIC underestimated unclosed ET, whereas SIMS overestimated closed ET, probably due to the assumption of well-watered conditions. In peak summer months, most models were in good agreement with closed ECET, with geeSEBAL and PT-JPL biased low. In September and October, when actual ET rates decline quickly, several models were biased high, except DisALEXI and geeSEBAL, which tracked closer to the unclosed values. The higher agreement of RSET with ECET during the peak summer period is encouraging, as this is the period of intensive irrigation and consumptive use of water through ET. A post hoc test showed that DisALEXI, geeSEBAL and SSEBop had mean monthly ET values that were statistically different (as underestimation) from the mean closed ECET. The mean aggregated growing season ET for all models were no different from the mean closed ECET (Supplementary Tables 8 and 9).
The monthly climatologies derived at flux sites were upscaled using data from all cropland pixels over the full OpenET domain (Extended Data Fig. 6). We found similar seasonal patterns and relative model biases to those identified at the flux sites-giving confidence in the representativeness of the ECET comparisons.

Impact of sampling interval on model performance

Model accuracy often improves with temporal aggregation interval due to cancellation of errors . In croplands, the accuracy metrics for the OpenET ensemble improved as the aggregation period increased
Fig. 2 | Modelled versus observed monthly ET at cropland sites. Monthly comparison of all paired OpenET ensemble members ET versus closed flux tower from all cropland stations for all months of record. Included for each model is the result of the least square linear regression model forced through the origin and .
from daily (overpass dates) to monthly to growing season to annual periods (Supplementary Tables 2-6). Daily ensemble results for the combined cropland sites showed a MAE of , and RMSE of of the mean ECET. At this timescale there is increased uncertainty both in the ECET data due to variability in micrometeorological conditions and energy balance closure, and remotely sensed ET due to potential cloud contamination and errors in footprint representation. These ensemble uncertainties are reduced when integrating to monthly (MAE of and RMSE of of ECET), growing season (MAE of and
RMSE of 15.5% of ECET) and water yer (MAE of 11.3% and RMSE of 12.3% of ECET) timescales. Fortunately, during growing season periods we found lower energy balance closure error in EC data and there is less cloud cover in satellite data in the western United States as compared with the non-growing period. During the summer, the daily ensemble normalized MAE (NMAE) on overpass dates was typically between 5% and (Supplementary Fig. 1), and monthly 7% and 20% (Fig. 4). We expect custom aggregation periods between 2 and 15 days to have similar or slightly improved accuracy to daily results that vary seasonally; subweekly to bi-weekly RSET may be of greatest use for irrigation scheduling .

Performance among annual and perennial crops

Annual crops, including wheat, corn, soy, rice and others, make up the majority (80%) of cropland sites in the OpenET ECET dataset (Supplementary Table 1). Compared with perennial crops, annual crops tend to have shorter canopies and more homogeneous cover at peakgrowth stage. The annual crop sites in the OpenET flux dataset are predominantly irrigated, and are distributed across a range of climatic zones, with higher density in regions such as Mediterranean and semi-arid Central Valley, California, and humid continental regions in the High Plains and the Mississippi Alluvial Plain (Fig.1).
For annual crops, each of the RSET models in the OpenET ensemble exhibited small bias and high levels of accuracy and precision (Table 1). Similar to all crop types combined, the ensemble value for annual crops outperformed individual models in terms of MAE ( 15.3 mm per month or 17.9% of mean ECET), RMSE ( 19.7 mm per month or 23.2% of mean ECET) and (0.9). Of the RSET models, eeMETRIC and PT-JPL exhibited the lowest magnitude of MBE, with PT-JPL and SIMS yielding the highest accuracy in terms of MAE and RMSE.
Dividing annual crops into C3 and C4 subclasses, we find the seasonal patterns and magnitudes of ensemble MAE are similar throughout the year (Fig. 4). NMAE in general reflects the inverse of the characteristic water use curve for each class, with C3 crops exhibiting a broader seasonal curve than C4 and therefore lower NMAE early and late in the season. While the higher NMAE values observed outside the growing season for all crop types (Fig. 4) are more indicative of low ET rates than of meaningful modelling error characteristics, cool-season errors may be generally inflated by higher cloud cover, increasing the time interval between cloud-free satellite retrievals. Improving satellite imaging frequency, as well as ET time integration and gap-filling techniques, should help to increase OpenET accuracy during the nongrowing season (Discussion).
Another class of interest is woody perennials, which are high-value crops and pose distinct modelling challenges. High-quality eddy flux ET data were available for three vineyards, three nut tree orchards and one fruit orchard, all located in California . Vineyards and orchards have taller and more highly structured canopies, often with inter-row cover crops, and vineyards are often deficit irrigated. These qualities lead to shadowing and mixed pixel effects in remote sensing at the level, and the need for sensitivity to small changes in vine stress to inform deficit irrigation applications is a unique modelling requirement.
RSET model performance in the vineyard sites sampled was strong and consistent across models. The ensemble accuracy exceeded that for annual crops (Table 1 and Fig. 4), with lower bias (slope of 1.02 and MBE of 5.3 mm per month) and lower MAE and RMSE ( 13.7 and 16.2 mm per month, respectively, or 12.2% and 14.5% of the mean monthly ECET) and of 0.90. DisALEXI performed similarly or better than the ensemble at the vineyard flux sites, perhaps due to its two-source approach towards partitioning temperature fluxes between the substrate (inter-row) and canopy.
Performance was more varied across ensemble members for the orchards than for other broad crop types, and biases were more negative. This could be related to shadowing effects in the taller and more
Table 2 | Summary statistics between modelled and observed monthly ET for cropland sites grouped by climate zone
Land cover type Statistic Ensemble DisALEXI eeMETRIC geeSEBAL PT-JPL SIMS SSEBop sites data points
Bsk + Bsh (cold and hot semi-arid steppe), mean station ET of 133 (mm per month) Slope 0.9 0.85 0.91 0.82 0.88 0.97 1 11 246
MBE (mm) -6.94 (-5.2%) -15.17 (-11.4%) -4.65 (-3.5%) -18.09 (-13.6%) -7.55 (-5.7%) 2.71 (2.0%) 2.89 (2.2%) 11 246
MAE (mm) 20.74 (15.6%) 26.93 (20.3%) 26.94 (20.3%) 30.31 (22.8%) 25.7 (19.3%) 22.99 (17.3%) 24.23 (18.2%) 11 246
RMSE (mm) 26.38 (19.8%) 34.55 (26.0%) 33.4 (25.1%) 38.0 (28.6%) 32.49 (24.4%) 28.97 (21.8%) 30.99 (23.3%) 11 246
0.89 0.81 0.8 0.8 0.84 0.84 0.84 11 246
Bwh + Bwk (hot and cold desert), mean station ET of 110 (mm per month) Slope 0.91 0.85 1.02 0.92 0.86 0.92 0.88 10 53
MBE (mm) -6.78 (-6.1%) -15.63 (-14.2%) 9.63 (8.7%) -8.19 (-7.4%) -7.77 (-7.0%) -3.2 (-2.9%) -12.34 (-11.2%) 7 49
MAE (mm) 13.24 (12.0%) 21.21 (19.2%) 18.92 (17.1%) 19.13 (17.3%) 19.21 (17.4%) 13.62 (12.3%) 19.59 (17.8%) 7 49
RMSE (mm) 17.02 (15.4%) 25.78 (23.4%) 23.92 (21.7%) 23.51 (21.3%) 23.81 (21.6%) 16.07 (14.6%) 22.95 (20.8%) 7 49
0.91 0.85 0.88 0.87 0.83 0.94 0.89 10 53
Cfa (humid subtropical), mean station ET of 75 (mm per month) Slope 1 1.03 1.03 0.99 0.93 1.15 0.91 11 232
MBE (mm) 2.15 (2.9%) 4.49 (6.0%) 3.65 (4.9%) 0.97 (1.3%) -0.88 (-1.2%) 14.98 (19.9%) -3.84 (-5.1%) 8 228
MAE (mm) 17.51 (23.3%) 20.17 (26.8%) 24.01 (31.9%) 20.06 (26.7%) 17.79 (23.6%) 22.71 (30.2%) 21.97 (29.2%) 8 228
RMSE (mm) 23.76 (31.6%) 26.18 (34.8%) 31.88 (42.4%) 28.39 (37.7%) 23.62 (31.4%) 30.23 (40.2%) 28.76 (38.2%) 8 228
0.75 0.72 0.62 0.69 0.76 0.72 0.64 11 232
Csa + Csb (hot- and warm-summer Mediterranean), mean station ET of 94 (mm per month) Slope 0.95 0.96 0.99 0.81 1.01 0.93 0.99 8 292
MBE (mm) -4.37 (-4.7%) -6.32 (-6.7%) -1.72 (-1.8%) -20.92 (-22.3%) 7.37 (7.9%) -1.34 (-1.4%) -2.9 (-3.1%) 8 292
MAE (mm) 13.32 (14.2%) 18.65 (19.9%) 17.14 (18.3%) 25.9 (27.6%) 17.17 (18.3%) 14.04 (15.0%) 23.94 (25.6%) 8 292
RMSE (mm) 16.52 (17.6%) 22.75 (24.3%) 21.48 (22.9%) 31.31 (33.4%) 21.97 (23.5%) 18.27 (19.5%) 27.92 (29.8%) 8 292
0.93 0.87 0.88 0.85 0.87 0.88 0.84 8 292
Dfa + Dfb (hot- and warm-summer humid continental), mean station ET of 67 (mm per month) Slope 0.9 0.93 0.94 0.86 0.87 1.02 0.9 13 829
MBE (mm) -8.0 (-11.9%) -8.07 (-12.0%) -6.74 (-10.0%) -10.55 (-15.7%) -5.72 (-8.5%) 4.88 (7.3%) -13.58 (-20.2%) 10 823
MAE (mm) 13.8 (20.5%) 15.68 (23.3%) 18.93 (28.2%) 17.88 (26.6%) 13.67 (20.3%) 15.35 (22.8%) 21.09 (31.4%) 10 823
RMSE (mm) 17.85 (26.6%) 20.36 (30.3%) 24.1 (35.8%) 23.36 (34.7%) 18.89 (28.1%) 19.94 (29.7%) 25.9 (38.5%) 10 823
0.91 0.9 0.83 0.86 0.89 0.86 0.86 13 829
Mean monthly summary statistics for comparisons between OpenET ensemble members ET and closed flux tower monthly ET for agricultural sites grouped by Köppen-Geiger climate
weighted mean closed flux tower ET.
strongly clumped canopies, particularly for models that are strongly dependent on LST inputs. The ensemble value had a negative bias with mean slope of per month, MAE 21.2 mm per month ( of ECET) and RMSE 27.9 mm per month (22.1% of ECET), and an of 0.91. SSEBop and SIMS had the least bias in terms of slope and MBE, and SSEBop and DisALEXI had the lowest error in terms of MAE and RMSE (Table1). While MAE in orchards is high mid-season, the normalized values are similar to those of annual crops (Fig. 4).

Variation of model performance across climate regions

To investigate variations in OpenET performance over different climates, cropland accuracy metrics were grouped by the Köppen-Geiger climate zones of the flux sites (Fig.1). Zones with fewer than five flux stations were omitted as a conservative measure, and some zones were lumped on the basis of secondary climate classifications (for example, hot- and warm-summer Mediterranean zones). Each resulting group had 7-13 flux stations used for calculation of accuracy statistics.
Overall, the OpenET ensemble had better agreement with ECET at crop sites in water-scarce, semi-arid to arid regions (Mediterranean and desert zones in the Southwest) as compared with humid zones (Table 2 and Supplementary Fig. 2). Irrigation is more prevalent in semi-arid to arid regions, and crop ET tends to be closer to potential ET rates and is more accurately modelled in some RSET modelling frameworks. High accuracy of models in semi-arid and arid regions is advantageous, given the high priority of water resource sustainability and management challenges in these regions.
Among the zones considered, the OpenET ensemble value was most accurate for crop sites in Mediterranean zones, with MAE of 13.3 and RMSE of 16.5 mm per month ( and of the mean ECET), with the ensemble outperforming individual members. Of the individual models, SIMS showed the best agreement with ECET in these regions, suggesting well-watered conditions for most sites or possible influence of adjacent non-irrigated areas on SEB models. Similarly, in arid sites (hot and cold desert), SIMS had the lowest MAE and RMSE (Table 2). During the growing season periods when the majority of irrigation is applied, the ensemble’s monthly NMAE was consistently below 10% for cropland sites in Mediterranean climates (Supplementary Fig. 2).
Model performance in the subhumid and humid continental regions of the Midwest and Central Plains was similar to that in the Mediterranean climate zone, again with the ensemble outperforming individual models in terms of collective statistics (Table 2 and Supplementary Fig. 2). Errors were higher at the humid subtropical sites, with SIMS tending to overestimate ET with a slope of 1.15 and normalized MBE of , indicating ET is less well correlated with vegetation density in this region, and that irrigation practices may result in intermittent vegetation water stress. Hypotheses for increased RSET error in humid regions and paths for improvement are proposed in Discussion.

Performance in natural ecosystems

Most of the flux stations (61%) used in the intercomparison were in non-agricultural sites, including shrublands, grasslands, mixed forests, conifer forests, and wetlands or riparian areas (Fig.1) . The SIMS model
is currently not designed for and implemented in non-agricultural land-cover types; for these pixels, the ensemble consists of five models with the possibility of removing a single outlier (Methods). Systematic model error and variability for non-agricultural sites was higher than cropland sites (Fig. 5).
Most models exhibited a high bias in wetland/riparian sites, dominated by overprediction of ET during the spring (Extended Data Fig.5). SSEBop had higher accuracy in these sites than other models and the ensemble value (Supplementary Tables 2-4). For models that estimate all components of the SEB (DisALEXI, eeMETRIC and geeSEBAL), this bias could result from an underestimation of the substrate (water) heat storage term in the spring before the vegetation canopy develops . These errors can potentially be mitigated in the future through accurate classification of inundated land areas.
Natural ecosystems under high water stress, such as shrublands and grasslands in desert and semi-arid steppe climates in the western United States, showed the highest variability and error with respect to ECET (Fig. 5 and Supplementary Tables 2-4). In these systems, ET can be a small fraction of available energy, and difficult to both measure on the ground and model using RSET approaches. Shrublands also tend to be more heterogeneous than cropland sites, and this can introduce additional uncertainty into model-measurement comparisons .Nevertheless, it is important to provide an evaluation of accuracy, both to benefit ET monitoring and land health assessments within shrub and grassland ecosystems, and to identify key areas for future research in RSET to reduce model error.
The Landsat-scale ET from OpenET also has applications in forested landscapes, as a predictor of forest health and mortality and as a metric of water yield response to forest management . In forested locations, most OpenET models overestimated ET, particularly at the evergreen flux sites sampled, yielding a slope for the ensemble value of 1.24 and MBE of 16.8 mm per month ( ). At these sites, eeMETRIC showed the least bias with a slope of 1.17 and an MBE of 10.8 mm per month (17.5%), while for MAE and RMSE, the ensemble value outperformed each individual model. At mixed forest sites, however, eeMETRIC and DisALEXI were in better agreement with ECET than was the ensemble.

Ensemble outlier removal and spatial inter-model variability

See Supplementary Discussion 2 for analysis and discussion of the MAD outlier removal approach that is used for computing the ensemble value, including spatial analysis of the occurrence of outliers and the long-term differences between each model’s seasonal ET and the ensemble value (Extended Data Figs. 7 and 8, Supplementary Figs. 3-9 and Supplementary Tables 9 and 10). Evidence suggests that the MAD approach showed accuracy metrics similar to other simple methods. Over 2016-2022, typically no model was identified as an outlier in cropland pixels; however, SIMS was about 10% more likely to be identified as an ensemble outlier, and it often gave the highest ET value, particularly in the Central Plains.

Discussion

ET is a critical driver and metric of ecosystem function, weather and climate, agricultural practices and water resource management. However, field-scale ET has previously been difficult to estimate at scale; therefore, ready access to high-resolution (spatially and temporally) ET data offers societal benefits to a variety of stakeholders . Using monthly ET data, water managers can develop more accurate water budgets in support of incentive-driven conservation programmes and innovative management and trading strategies. For policymakers, such data can improve water supply tracking, simplify regulatory compliance and promote the co-development of solutions with local communities. Crop producers may be able to improve the efficiency of irrigation practices in some instances, resulting in enhanced sustainability and reduced
Fig. 3 | Monthly climatology of paired modelled and observed ET for cropland sites. a, Monthly climatology of paired OpenET and flux tower from cropland sites. , The residual of monthly mean ET (model minus mean closed flux ET). Unclosed and closed labels refer to flux tower ET before and after energy balance closure correction. Dashed lines represent the closed flux ET mean plus two standard errors of the mean and unclosed flux ET mean minus two standard errors of the mean.
costs for water, fertilizer and energy. Supplementary Discussion 3 continues the conversation on incentives towards improving irrigation efficiency and how OpenET data can provide value in an RSET-based irrigation scheduling framework.
In addition to informing water management, OpenET has multiple research and modelling applications. Carbon and climate modelling can benefit from 30-m RSET data as a diagnostic indicator of ecosystem health and function response under a changing climate . RSET is being used to reduce summertime warm-dry bias in weather forecasting and climate models by improving the representation of ET from irrigated land , ET-soil moisture coupling and transpiration-evaporation partitioning . Hydrologic and land surface models at multiple scales can also benefit from high-resolution ET data, for example, as validation or forcing data in basins where streamflow measurements are not available to constrain the water budget .
Realizing the full potential benefits of RSET data for water resource and land management applications requires rigorous and reproducible accuracy assessment to inform practitioners on best use practices . The accuracy results we present here provide valuable constraints on model uncertainty based on broad crop type, climate region and timescale.
Average error in the OpenET ensemble value with respect to mean ECET in cropland sites for monthly, growing season and annual aggregated ET, ranged from 10% to 17% for MAE and 11% to 22% for RMSE. These errors are within accuracy levels of reported for supervised remote sensing techniques . They are also consistent with accuracy targets set by the OpenET user groups: 10-20% at a
Fig. 4 | Monthly MAE of the model ensemble for different croptypes. a, Monthly mean flux tower . b, OpenET ensemble MAE and MAE normalized by the mean flux tower (NMAE) using all paired modelmeasured data for cropland stations grouped by crop types. Annual crops that had a mixed history of rotation between C3 and C4 crop types, for example, corn-soy rotations, were not included in C3 or C4 results but were included in the combined grouping.
monthly timestep, and for daily ET data . These errors include uncertainties in ECET data, which are estimated to range from 10% to 30% depending on site characteristics and instrumentation design and maintenance .
These accuracy results may support advancements in water management applications that incorporate OpenET data. For croplands, all models except for SIMS had negative bias errors at the monthly timestep ( to ), with an MBE of for the ensemble ET value (SIMS MBE is ). Awareness of these bias errors when using these data for irrigation management applications may prevent unintentional deficit irrigation that can suppress crop yields and farm revenue . Cross-comparisons between the primarily reflectancebased SIMS and PT-JPL models and the LST-driven models may be useful for identifying periods of intentional or unintentional crop water stress and deficit irrigation. Reducing errors in the OpenET daily data is a high priority for advancing their utility for on-farm water management.
At local to regional scales, the reported uncertainties at monthly to annual timesteps should inform applications related to water balance, water accounting and water rights administration. Comparison of OpenET data aggregated at the scale of irrigation districts or watersheds against carefully constrained water balances offers one path to assessment of biases at larger scales. Particularly in administration of water rights, the current uncertainty in the OpenET data (for example, growing season ensemble NMAE of for croplands) must be recognized in evaluating consumptive water use, and OpenET data should only be used for this purpose in combination with other sources of information.
This study provides insights into potential pathways towards improving the accuracy of the individual models within the OpenET ensemble. Across both agricultural and some natural landscapes, most models underestimated cropland ET during the winter and spring, particularly the models that rely upon TIR measurements to compute ET. This underestimation may be related to loss of thermal contrast over an image, where differences between the hottest and coolest pixels are reduced relative to midsummer values, adding uncertainty to within-scene scaling approaches. It may also be related to misrepresentation of soil evaporation during extended wet periods, extended periods of cloudiness, and error in shared model inputs. In addition, treatment of effects of senesced standing vegetation and crop residue on SEB can impact model performance outside of the growing season. In terms of observational errors, the energy balance closure error and uncertainty in EC data are also amplified during periods outside of the growing season .
We found increased model error in croplands in humid climates as compared with drier regions. Again, lower temperature contrasts across humid landscapes may contribute to errors in TIR-based withinscene scaling models. A primary driver, however, is probably the relative paucity of clear-sky satellite retrievals and potential for error in LST due to undetected clouds. Improving temporal sampling of RSET model inputs will be a major focus of on-going development in OpenET, through future use of imagery from additional Landsat-like optical (Sentinel-2) and thermal (ECOSTRESS, VIIRS) sensors , and integration of future TIR observations from satellite missions currently in development by NASA, USGS and the European Space Agency. Methods for computing ET values between cloud-free satellite observations, currently based on linear interpolation of the ratio of ET to a reference flux, can also be improved. Approaches used in mapping and predicting vegetation phenology and dynamic time warping algorithms developed for signal processing applications offer promise for reducing large errors during periods of rapid vegetation change or extended cloud cover, which would contribute to reduced RMSE values across the model ensemble.
Examining results for specific crop classes, we found strong results for DisALEXI and SIMS over vineyards, and DisALEXI, SIMS and SSEBop over fruit and nut orchard sites-key targets for irrigation management in the Central Valley. Increasing the number of validation sites in orchards would help to address remaining modelling issues associated with this challenging canopy architecture. The USDA ARS-led Tree-crop Remote sensing of Evapotranspiration eXperiment (T-REX) is aimed at addressing this observational gap .
All models, to varying degrees, have room for notable improvement in computation of ET in natural ecosystems. For example, most models systematically underestimate ET in drier ecosystems such as grasslands and shrublands and overestimate ET in evergreen forests. Incorporation of high-frequency and high-resolution visible and nearinfrared data into the remote sensing models may improve their ability to capture phenological shifts particularly in arid/semi-arid regions, and agricultural systems in general . Improvement of gridded meteorological model inputs , land cover classification data and soils data may also lead to improved model performance in both natural ecosystems and in croplands. In particular, datasets compiled from agricultural weather stations and used to compute bias correction surfaces for reference ET could be re-evaluated to ensure reference surface compliance with the assumptions of the American Society of Civil Engineers Penman-Monteith equation .
Future OpenET accuracy evaluations will target primary causes of error in ground ET measurements and RSET methods. Specific factors to consider include local advective impacts on modelled and measured ET, EC energy budget closure, local thermal contrast, ET reduction in deficit irrigated or rainfed systems, potential biases in gridded meteorological inputs to RSET models, and accurate capture of ET over sparsely cultivated landscapes. Comparisons with other
Fig. 5 | Monthly modelled ensemble versus observed ET for sites grouped by land cover type. Monthly comparison of the paired monthly OpenET versus closed flux tower for each general land cover group. Included for each group is the result of the least square linear regression model and .
well-established spatially mapped ET products such as MOD16 or FLUXCOM may provide further insights for operational global ET mapping at field scales ( ). Comparisons against ET data computed from long-term water balance studies would help fill in gaps of spatial coverage in measured in situ ET across the western United States in hydrologically important but sparsely cultivated regions such as the Upper Colorado River Basin.

Conclusions

The OpenET platform provides spatially continuous ET data at resolution throughout the western United States. An intercomparison and accuracy assessment involved six satellite-based RSET models composing the current OpenET version, ensemble ET computed from the six models, and a well-documented benchmark eddy flux dataset from 152 stations located in the contiguous United States. Based on results from 59 cropland ET stations located in a variety of climatic regions, little systematic model bias was observed in croplands, and error metrics were within or near the targets set forth by OpenET partners including farmers, irrigation managers and water management agencies. The best accuracy metrics were associated with seasonal and annual timescales, and for crops in arid/semi-arid regions. The OpenET ensemble mean, with outlier removal, typically outperformed any individual model in terms of error statistics. Generally, no more than one model was identified as an outlier during growing season months over most agricultural regions in the western United States, and frequently no models were excluded. This finding highlights the substantial progress achieved so far in developing fully automated RSET modelling approaches that can be employed to map ET over large areas at field-scale resolution. The study identified paths for future targeted research and model improvement, and is intended to support the RSET research community in the development of increasingly robust and accurate RSET techniques. We are also hopeful that this assessment will provide added confidence to water resource managers, farmers, ranchers, scientists and other
potential users of OpenET due to the high rigour and transparency of methods that were employed.

Methods

Flux data processing and footprint sampling

We used a curated benchmark eddy flux-based ET dataset and tools for use in this and subsequent evaluations of OpenET RSET models . The rationale and decision-making steps for the collection and post-processing of flux data, as well as analyses of footprint sampling techniques and energy balance closure error within the dataset, are described in Volk et al. . Data processing techniques for gap-filling and correction for energy balance closure error were conducted using open-source Python tools that enhance data provenance and reproducibility. Data were also subject to qualitative, visual-based data screening and filtering . The final post-processed dataset consists of 161 stations, is public and includes daily and monthly ET and meteorological data, interactive graphics of such data for each station, and site information such as land use and Principal Investigator acknowledgements . We note that nine stations in the dataset were not included in the statistical results presented here because they had data coverage that did not overlap with the data that could be developed for all six OpenET models. For example, not all models could be implemented from satellite imagery recorded before 2001 (ref. 5). Figure 1 shows a map of the 152 stations used in this accuracy assessment as well as their land cover types and Köppen-Geiger climate zones, and Supplementary Table 1 provides additional metadata for each station.
Data for the majority (106) of the flux stations in this study were downloaded from the AmeriFlux website, last accessed on 27 October 2020, and the remaining stations were retrieved from a variety of sources and Principal Investigators from university partners, the US Geological Survey, the US Department of Agriculture and oth . In addition to EC systems, four precision weighing lysimeters measuring cropland ET in Texas and seven high-quality Bowen Ratio
instrumented sites, which measure ET in predominantly phreatophyte shrublands in Nevada , were included in the dataset. Gapfilling of initial half-hourly fluxes of the four main energy balance components-latent, sensible and soil heat flux, and net radiationwas conducted using linear interpolation where gaps up to 2 h during the daytime or 4 h during nighttime were interpolated. If a given period still contained gaps then the daily average was not calculated and the daily flux value was left as a gap. After this initial gap-filling, fluxes were averaged to daily periods and energy balance closure correction was applied following the daily energy balance ratio approach defined by FLUXNET2015/ONEFlux . The corrected daily latent heat flux, which is the energy consumed through ET, was used to calculate ET with an adjustment to the latent heat of vapourization for air temperature . This closure-adjusted value is referred to as closed flux ET or measured ET in the main text and all statistical measures reported for OpenET models were against the energy balance corrected ET data. Daily ET gaps were subsequently filled using gridMET fraction of reference ET and gridMET grass reference . To exclude flux stations with higher data uncertainty, only stations with mean daily energy balance closure of 0.75 or higher during the growing season and 0.6 or higher during the non-growing season were chosen for this intercomparison. Here, growing season periods were spatially mapped on the basis of a cumulative growing-degree-day and killing frost approach derived from long-term gridded climate data and are specific to each flux site . The final dataset is similar to the recent FLUXNET2015 (ref. 29) release consisting of high-quality eddy flux station data that were subject to similar processing and correction techniques. The largest difference between the two datasets, in terms of daily latent heat flux estimates, results from different gap-filling procedures, where our approach is considered to be simpler and more conservative .
Two approaches were used to estimate flux tower footprints or source area for tower pixel sampling of RSET imagery:(1) simple square ‘static’ pixel (Landsat 30 m ) grids of and drawn around station locations, and (2) two-dimensional, physically based flux source area estimations modelled using hourly meteorological data using the Kljun et al. approach, with hourly footprints converted to daily/ monthly average footprint rasters weighted by reference . The placement of the static grids was informed by high-resolution imagery to avoid inclusion of pixels of non-representative land cover (structures, roads and canals), and shifted slightly into the predominant wind direction as determined by long-term mean daytime windroses (built from data between 6:00 and 20:00 local time). Although the physically based and temporally dynamic footprints were preferred over the static footprints, only about half of the stations in the dataset had sufficient data for their production. Commonly, one or more input parameters to the Kljun et al. model, such as the standard deviation of the crosswind component of wind due to turbulence or friction velocity, was not available. A detailed description of parameter estimation, processing steps and the method used for creating weighted mean footprint images (using reference ET from NLDAS2 gridded weather data ) can be found in Volk et al. . We also conducted a rigorous comparison of the intersection between source areas from the static grids of different sizes and the temporally dynamic footprints. The major finding was that the larger grids tended to include substantially more of the dynamically defined footprint area than did the smaller grid sizes on average; however, the smaller grids tended to overlap with pixels that were deemed part of the dynamic footprint on a more consistent basis. Therefore, we decided to use the grids for pixel sampling at most flux sites where a dynamic footprint could not be generated, with exceptions for sites with heterogeneous surroundings or with non-representative land cover nearby the station. For these sites, we used or grids to avoid giving equal weight to pixels of potentially different land cover that lie near the perimeter of the typical actual footprint area .

Model data

The majority of the models that make up the OpenET ensemble are based on full or simplified implementations of the SEB approach. The SEB approach accounts for the energy used to transform liquid water in plants and soil into vapour that is released to the atmosphere. The SEB approach relies on satellite measurements of surface temperature and surface reflectance combined with other key land surface and weather variables to calculate components of the energy balance-net radiation, sensible heat flux, ground heat flux and latent heat flux. eeMETRIC , geeSEBAL and DisALEXI compute each component of the energy balance using optical (that is, short-wave) and thermal (that is, longwave) data, whereas SSEBop and PT-JPL are simplified approaches in which certain components of the energy balance are not calculated, or are calculated using a set of simplifying assumptions. SIMS relies on surface reflectance data, crop type information and a gridded soil water balance model to compute ET as a function of canopy density using a crop coefficient approach for agricultural lands.
The Google Earth Engine Python application programming interface was used to develop a workflow for sampling OpenET RSET model data at ET flux sites. Sampling of the daily and monthly RSET model data was performed at each site using a set of static ( and/or ) and/or dynamic flux source-area footprints. Conditions for each of the extraction methods using static footprints were as follows: (1) daily ET from eeMETRIC, SIMS and SSEBop for sites outside of California was calculated as the product of the mean daily fraction of grass reference ET (EToF) produced by the models and the mean daily bias-corrected gridMET grass reference ET (ETo) (repeated for sites within California using daily CIMIS ETo, where CIMIS is more commonly used and depended upon in California); (2) daily ET from PT-JPL, geeSEBAL, and ALEXI/DisALEXI for all sites was computed as the spatial average of daily ET pixels produced by the models; (3) monthly ET from all RSET models for sites outside of California were calculated as the product of the mean monthly EToF and the mean monthly gridMET ETo (repeated for sites within California using the monthly CIMIS ETo). The process of extrapolating instantaneous data (time of overpass) to daily ET is an internal model calculation and differs for each model, and we refer readers to the individual model documentations for details as well as Melton et al. . Daily Landsat image pixels with cloud contamination are flagged on the basis of the CFMask derived indicators in the pixel quality assurance band (QA_PIXEL) and those pixels are not considered. When computing monthly ET, all missing or masked daily ET pixels are computed by linearly interpolating between the nearest unmasked (cloud free) pixels in time within days.
Conditions for each of the extraction methods using dynamic footprints were as follows:
(1) daily ET from eeMETRIC, SIMS and SSEBop for sites outside of California was calculated by first multiplying the sampled daily EToF pixels produced by the models in the footprint by each daily flux footprint weight to obtain daily weighted EToF pixels, and summing all daily weighted EToF pixels to obtain mean daily weighted EToF, normalizing the mean daily weighted EToF by the sum of weights to account for times when the sum of weights did not equal 1 (for example, caused by cloud masking of pixels), and then multiplying the mean daily weighted EToF by the mean daily bias corrected gridMET ETo (replaced for sites within California using the daily CIMIS ETo);
(2) daily ET from PT-JPL, geeSEBAL and ALEXI/DisALEXI for all sites was calculated by multiplying the daily ET pixels by the daily flux footprint weights to obtain daily weighted ET pixels, summing all daily weighted ET pixels to obtain mean daily weighted ET, and then normalizing the mean daily weighted ET by the sum of weights, and
(3) monthly ET from all RSET models for sites outside of California was calculated by first multiplying the monthly EToF pixels by
the monthly flux footprint weights to obtain monthly weighted EToF pixels, summing all monthly weighted EToF pixels to obtain mean monthly weighted EToF, normalizing the mean monthly weighted EToF by the sum of weights, and then multiplying the mean monthly weighted EToF by the mean monthly bias-corrected gridMET ETo (replaced for sites within California using the monthly CIMIS ETo).
Additional processing was required after extracting the daily ET when duplicate days of data were extracted at select sites due to overlapping Landsat paths. Occasionally a site would lie within the footprints of two overlapping Landsat scenes, resulting in more than one ET value on a given overpass date. To obtain single daily ET values for the site, the daily weighted mean ET for each day was computed using the pixel count (that is, number of pixels used when deriving the respective spatial mean ET value) as the weight. ET pixel counts were occasionally less than the grid/footprint total because of the removal of poor-quality pixels (for example, cloud masking).

Ensemble computation

The ensemble mean of the six OpenET models was computed after removing up to two outlier models based on the MAD , a robust measure of spread that is suitable for small samples. The outlier removal occurs at the pixel level for each ET image generated. To identify outliers for a single scene, first the median value and the MAD from the median is computed as
where is the ET value for model and is the full set of all six model’s ET estimates. Here, is a scalar set to 1.483, and it was derived on the basis of the assumption of normality of the sample population . This approach is sometimes referred to as the MADe rule, where . The MAD value is typically scaled by or 3 on the basis of a subjective assessment of the data, which is then used to create a band around the median:
Model estimates that fall outside the band are deemed as outliers, and up to two outliers (those furthest from the median) are removed from the set of model estimates before taking the ensemble mean.
Due to the tendency for some OpenET models to predict zero ET or even negative ET rates in some arid regions during dry periods we modified the above approach for these scenarios. Specifically, when the ensemble median estimate is zero but at least one model predicts a positive ET rate, the ensemble mean is taken to include that value without any prior outlier removal. In these cases, the outlier removal would result in removing the model estimates that are positive and although actual ET may be quite negligible, a zero estimate is not considered to be physically realistic. However, in these scenarios, because the majority of models may predict zero, the ensemble mean will also be highly skewed towards zero making this a conservative measure to prevent zero ensemble estimates.

Statistical analyses

Key summary statistics including the least squares linear regression slope forced through the origin (slope) as well as linear regression with an intercept (Supplementary Table 7), MBE, MAE, RMSE and the coefficient of determination ( ) were computed using paired observations between OpenET model ET estimates and post-processed and corrected flux ET estimates . Daily accuracy statistics were not compared against any gap-filled station ET data, and monthly statistics only used station ET with 5 or fewer gap-filled days per month. Growing season and annual evaluations used paired monthly data and did not include any periods with monthly gaps. Also, the number
of paired observations was always the same among models for all statistical analyses.
All statistics were calculated on a site-by-site basis using paired model-measured ET using the Python Numpy package version 1.17.2 (ref. 63). For linear regression, the Numpy linalg.Istsq algorithm was used, and it applies the least squares approach. We used the modelled ET as the dependent variable and the measured ET as the independent variable.
The MBE was calculated as
where is the observed is the model predicted ET and is the total number of paired model-measured ET data points.
The MAE was calculated as
and the RMSE was calculated as
Here, values were calculated as the square of the Pearson correlation coefficient, which was calculated from paired model-measurement ET data using the Python statsmodels package, version 0.12.1 (ref. 64).
For grouping statistics by land cover or climate zone we used two methods: (1) for the computation of linear regression and all data from each ground observation in a group (for example, monthly paired model-station ET estimates for annual crop stations) were pooled together before computing a single statistic per model; and (2) MBE, MAE and RMSE were computed separately for each ground station, and then a weighted mean was taken. Grouped statistics were weighted by the square root of the number of paired observations per station ( ); the rationale is to avoid giving too much weight to stations with excessively long data records while also not giving equal weight to stations with short data records . We also imposed data length requirements for in situ ET stations: to be included in daily grouped mean statistics we required stations to have a minimum of six paired station-model data points, and a minimum of three paired observations for inclusion in monthly grouped mean statistics. We note that Melton et al. presented similar statistical metrics from a subset of cropland sites used in this study, and in that study, the linear regression slope and metrics did incorporate weighting, which we deemed inappropriate or unnecessary in this study. For congruency, the statistics computed in the same manner as in Melton et al. are provided in Supplementary Table 12.
A post hoc Tukey test, also known as the honestly significant difference test, was used to compare multiple mean ET estimates from each model, the ensemble mean, and from the mean of the unclosed and closed flux ET data. The test was applied using all paired data from cropland stations, including for crop subgroups: annual crops, orchards and vineyards, at daily, monthly, growing season and annual timescales. The family-wise error rate was set to 0.05 and the test was performed using the Python statsmodels package, version 0.12.1 (ref. 64).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

The in situ measured ET data analysed during the current study are available in the Zenodo repository, with identifier https://doi. org/10.5281/zenodo.7636781. The OpenET model ET data analysed
during the current study are available in the Zenodo repository, with identifier https://doi.org/10.5281/zenodo.10119477.

Code availability

The code used to post-process eddy flux tower data for the current study is publicly available on GitHub (https://github.com/Open-ET/ flux-data-qaqc). The code used to generate flux footprints for the current study is publicly available on GitHub (https://github.com/ Open-ET/flux-data-footprint).

References

  1. Fisher, J. B. et al. The future of evapotranspiration: global requirements for ecosystem functioning, carbon and climate feedbacks, agricultural management, and water resources. Water Resour. Res. 53, 2618-2626 (2017).
  2. Dieter, C. A. et al. Estimated use of water in the United States in 2015. Circular 1411 https://pubs.usgs.gov/publication/cir1441 (2018).
  3. Cook, B. I., Ault, T. R. & Smerdon, J. E. Unprecedented 21st century drought risk in the American Southwest and Central Plains. Sci. Adv. 1, e1400082 (2015).
  4. Liu, P.-W. et al. Groundwater depletion in California’s Central Valley accelerates during megadrought. Nat. Commun. 13, 7825 (2022).
  5. Melton, F. S. et al. OpenET: filling a critical data gap in water management for the western United States. J. Am. Water Resour. Assoc. 58, 971-994 (2022).
  6. Chen, J. M. & Liu, J. Evolution of evapotranspiration models using thermal and shortwave remote sensing data. Remote Sens. Environ. 237, 111594 (2020).
  7. Anderson, M. et al. Field-scale assessment of land and water use change over the California Delta using remote sensing. Remote Sens. 10, 889 (2018).
  8. Allen, R. G., Tasumi, M. & Trezza, R. Satellite-based energy balance for mapping evapotranspiration with internalized calibration (METRIC)-Model. J. Irrig. Drain. Eng. 133, 380-394 (2007).
  9. Laipelt, L. et al. Long-term monitoring of evapotranspiration using the SEBAL algorithm and Google Earth Engine cloud computing. ISPRS J. Photogramm. Remote Sens. 178, 81-96 (2021).
  10. Fisher, J. B., Tu, K. P. & Baldocchi, D. D. Global estimates of the land-atmosphere water flux based on monthly AVHRR and ISLSCP-II data, validated at 16 FLUXNET sites. Remote Sens. Environ. 112, 901-919 (2008).
  11. Pereira, L. S. et al. Prediction of crop coefficients from fraction of ground cover and height. Background and validation using ground and remote sensing data. Agric. Water Manag. 241, 106197 (2020).
  12. Melton, F. S. et al. Satellite irrigation management support with the terrestrial observation and prediction system: a framework for integration of satellite and surface observations to support improvements in agricultural water resource management. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 5, 1709-1721 (2012).
  13. Senay, G. B. et al. Improving the operational simplified surface energy balance evapotranspiration model using the forcing and normalizing operation. Remote Sens. 15, 260 (2023).
  14. Gorelick, N. et al. Google Earth Engine: planetary-scale geospatial analysis for everyone. Remote Sens. Environ. 202, 18-27 (2017).
  15. Allen, R. G. et al. Satellite-based energy balance for mapping evapotranspiration with internalized calibration (METRIC)Applications. J. Irrig. Drain. Eng. 133, 395-406 (2007).
  16. Knipper, K. R. et al. Using high-spatiotemporal thermal satellite ET retrievals for operational water use and stress monitoring in a California vineyard. Remote Sens. 11, 2124 (2019).
  17. Senay, G. B., Friedrichs, M., Singh, R. K. & Velpuri, N. M. Evaluating Landsat 8 evapotranspiration for water use mapping in the Colorado River Basin. Remote Sens. Environ. 185, 171-185 (2016).
  18. Foster, T., Mieno, T. & Brozović, N. Satellite-based monitoring of irrigation water use: assessing measurement errors and their implications for agricultural water management policy. Water Resour. Res. 56, e2020WRO28378 (2020).
  19. Volk, J. M. et al. Development of a benchmark eddy flux evapotranspiration dataset for evaluation of satellite-driven evapotranspiration models over the CONUS. Agric. For. Meteorol. 331, 109307 (2023).
  20. Volk, J. M. et al. Post-processed data and graphical tools for a CONUS-wide eddy flux evapotranspiration dataset. Data Brief https://doi.org/10.1016/j.dib.2023.109274 (2023).
  21. Baldocchi, D. Measuring fluxes of trace gases and energy between ecosystems and the atmosphere-the state and future of the eddy covariance method. Glob. Change Biol. 20, 3600-3609 (2014).
  22. Baldocchi, D. et al. FLUXNET: a new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide, water vapor, and energy flux densities. Bull. Am. Meteorol. Soc. 82, 2415-2434 (2001).
  23. Hampel, F. R. The influence curve and its role in robust estimation. J. Am. Stat. Assoc. 69, 383-393 (1974).
  24. Leys, C., Ley, C., Klein, O., Bernard, P. & Licata, L. Detecting outliers: do not use standard deviation around the mean, use absolute deviation around the median. J. Exp. Soc. Psychol. 49, 764-766 (2013).
  25. Thompson, P. D. How to improve accuracy by combining independent forecasts. Mon. Weather Rev. 105, 228-229 (1977).
  26. Kirtman, B. P. et al. The North American multimodel ensemble: phase-1 seasonal-to-interannual prediction; phase-2 toward developing intraseasonal prediction. Bull. Am. Meteorol. Soc. 95, 585-601 (2014).
  27. Bai, Y. et al. On the use of machine learning based ensemble approaches to improve evapotranspiration estimates from croplands across a wide environmental gradient. Agric. For. Meteorol. 298, 108308 (2021).
  28. Novick, K. A. et al. The AmeriFlux network: a coalition of the willing. Agric. For. Meteorol. 249, 444-456 (2018).
  29. Pastorello, G. et al. The FLUXNET2O15 dataset and the ONEFlux processing pipeline for eddy covariance data. Sci. Data 7, 1-27 (2020).
  30. Mauder, M., Foken, T. & Cuxart, J. Surface-energy-balance closure over land: a review. Bound. Layer Meteorol. 177, 395-426 (2020).
  31. Ingwersen, J., Imukova, K., Högy, P. & Streck, T. On the use of the post-closure methods uncertainty band to evaluate the performance of land surface models against eddy covariance flux data. Biogeosciences 12, 2311-2326 (2015).
  32. Knipper, K. R. et al. Evapotranspiration estimates derived using thermal-based satellite remote sensing and data fusion for irrigation management in California vineyards. Irrig. Sci. 37, 431-449 (2019).
  33. Bambach, N. et al. Evapotranspiration uncertainty at micrometeorological scales: the impact of the eddy covariance energy imbalance and correction methods. Irrig. Sci. 40, 445-461 (2022).
  34. Rubel, F., Brugger, K., Haslinger, K. & Auer, I. The climate of the European Alps: shift of very high resolution Köppen-Geiger climate zones 1800-2100. Meteorol. Z. 26, 115-125 (2017).
  35. Yang, Y. et al. Studying drought-induced forest mortality using high spatiotemporal resolution evapotranspiration data from thermal satellite imaging. Remote Sens. Environ. 265, 112640 (2021).
  36. Isaacson, B. N., Yang, Y., Anderson, M. C., Clark, K. L. & Grabosky, J. C. The effects of forest composition and management on evapotranspiration in the New Jersey pinelands. Agric. For. Meteorol. 339, 109588 (2023).
  37. Qian, Y. et al. Neglecting irrigation contributes to the simulated summertime warm-and-dry bias in the central United States. Npj Clim. Atmos. Sci. 3, 31 (2020).
  38. Lei, F., Crow, W. T., Holmes, T. R., Hain, C. & Anderson, M. C. Global investigation of soil moisture and latent heat flux coupling strength. Water Resour. Res. 54, 8196-8215 (2018).
  39. Dong, J., Lei, F. & Crow, W. T. Land transpiration-evaporation partitioning errors responsible for modeled summertime warm bias in the central United States. Nat. Commun. 13, 336 (2022).
  40. Abolafia-Rosenzweig, R., Pan, M., Zeng, J. & Livneh, B. Remotely sensed ensembles of the terrestrial water budget over major global river basins: an assessment of three closure techniques. Remote Sens. Environ. 252, 112191 (2021).
  41. Wang, Q. et al. Land surface models significantly underestimate the impact of land-use changes on global evapotranspiration. Environ. Res. Lett. 16, 124047 (2021).
  42. Allen, R. G., Pereira, L. S., Howell, T. A. & Jensen, M. E. Evapotranspiration information reporting: I. Factors governing measurement accuracy. Agric. Water Manag. 98, 899-920 (2011).
  43. Adu, M. O., Yawson, D. O., Armah, F. A., Asare, P. A. & Frimpong, K. A. Meta-analysis of crop yields of full, deficit, and partial root-zone drying irrigation. Agric. Water Manag. 197, 79-90 (2018).
  44. Xue, J. et al. Improving the spatiotemporal resolution of remotely sensed ET information for water management through Landsat, Sentinel-2, ECOSTRESS and VIIRS data fusion. Irrig. Sci. 40, 609-634 (2022).
  45. Gao, F. & Zhang, X. Mapping crop phenology in near real-time using satellite remote sensing: challenges and opportunities. J. Remote Sens. 2021, 8379391 (2021).
  46. Müller, M. Dynamic time warping. in Information Retrieval for Music and Motion. 69-84 (Springer, 2007).
  47. Bambach, N. et al. The Tree-crop Remote sensing of Evapotranspiration eXperiment (T-REX): a science-based path for sustainable water management and climate mitigation. Bull. Am. Meteorol. Soc. In the press (2023).
  48. Fisher, J. B. Hydrosat: towards daily, field-scale, global evapotranspiration from space. (2022).
  49. Polhamus, A., Fisher, J. B. & Tu, K. P. What controls the error structure in evapotranspiration models? Agric. For. Meteorol. 169, 12-24 (2013).
  50. Blankenau, P. A., Kilic, A. & Allen, R. An evaluation of gridded weather data sets for the purpose of estimating reference evapotranspiration in the United States. Agric. Water Manag. 242, 106376 (2020).
  51. Doherty, C. T. et al. Effects of meteorological and land surface modeling uncertainty on errors in winegrape ET calculated with SIMS. Irrig. Sci. 40, 515-530 (2022).
  52. Purdy, A., Fisher, J., Goulden, M. & Famiglietti, J. Ground heat flux: an analytical review of 6 models evaluated at 88 sites and globally. J. Geophys. Res. Biogeosci. 121, 3045-3059 (2016).
  53. Allen, R. G. et al. A recommendation on standardized surface resistance for hourly calculation of reference ETo by the FAO56 Penman-Monteith method. Agric. Water Manag. 81, 1-22 (2006).
  54. Jung, M. et al. The FLUXCOM ensemble of global landatmosphere energy fluxes. Sci. Data 6, 74 (2019).
  55. Reitz, M., Senay, G. B. & Sanford, W. E. Combining remote sensing and water-balance evapotranspiration estimates for the conterminous United States. Remote Sens. 9, 1181 (2017).
  56. Volk, J. et al. flux-data-qaqc: a Python package for energy balance closure and post-processing of eddy flux. Data. 6, 1-5 (2021).
  57. Evett, S. R. et al. The Bushland weighing lysimeters: a quarter century of crop ET investigations to advance sustainable irrigation. Trans. ASABE 59, 163-179 (2016).
  58. Abatzoglou, J. T. Development of gridded surface meteorological data for ecological applications and modelling. Int. J. Climatol. 33, 121-131 (2013).
  59. Kljun, N., Calanca, P., Rotach, M. W. & Schmid, H. P. A simple twodimensional parameterisation for Flux Footprint Prediction (FFP). Geosci. Model Dev. 8, 3695-3713 (2015).
  60. Xia, Y. et al. Continental-scale water and energy flux analysis and validation for the North American Land Data Assimilation System project phase 2 (NLDAS-2): 1. Intercomparison and application of model products. J. Geophys. Res. Atmos. 117, DO31O9 (2012).
  61. Foga, S. et al. Cloud detection algorithm comparison and validation for operational Landsat data products. Remote Sens. Environ. 194, 379-390 (2017).
  62. Rousseeuw, P. J. & Croux, C. Alternatives to the median absolute deviation. J. Am. Stat. Assoc. 88, 1273-1283 (1993).
  63. Harris, C. R. et al. Array programming with NumPy. Nature 585, 357-362 (2020).
  64. Seabold, S. & Perktold, J. Statsmodels: econometric and statistical modeling with Python. In Proc. 9th Python in Science Conference vol. 57 10-25080 (SciPy, 2010).
  65. Obrecht, N. A. Sample size weighting follows a curvilinear function. J. Exp. Psychol. Learn. Mem. Cogn. 45, 614 (2019).

Acknowledgements

OpenET data used in this study was produced on Google Earth Engine and we gratefully acknowledge Google, Inc. for the computing support and resources used to produce and process these data. Work on this analysis was supported by the Walton Family Foundation; Lyda Hill Philanthropies; National Aeronautics and Space Administration (NASA) Applied Science Program (grant NNX17AF53G, J.L.H.; NNX12AD05A, F.M.); United States Geological Survey (USGS) NASA Landsat Science Team (grant number 140G0118C0007, J.L.H.); USGS Cooperative Ecosystem Studies Units (CESU) (grant G23ACOO568, A.P.); USGS Water Resources Research Institute (grant G22AC00584-00, J.L.H.), NASA Western Water Applications Office (grant 1669431, J.L.H.; 80NSSC23K0836, C.P.); California State University Agricultural Research Institute (grant number 21-01-106, F.M.); National Institute of Food and Agriculture McIntire Stennis (grant MISZ-721160, Y.Y.); Desert Research Institute Maki Endowment; and the Idaho Agricultural Experiment Station and Nebraska Agricultural Experiment Station. In-kind support is provided by partners in the agricultural and water management communities, Environmental Defense Fund and Google Earth Engine. We acknowledge and thank the long-term data collection efforts by the AmeriFlux program, the US Department of Agriculture (USDA) Agricultural Research Service, the USGS, and the USDA National Agricultural Statistics Service. The DOIs for the AmeriFlux and USGS stations used in this study include the following: https:// doi.org/10.17190/AMF/1436327, https://doi.org/10.17190/AMF/1436328, https://doi.org/10.17190/AMF/1418680, https://doi.org/10.17190/ AMF/1246137, https://doi.org/10.17190/AMF/1246025, https://doi. org/10.17190/AMF/1246026, https://doi.org/10.17190/AMF/1246027, https://doi.org/10.17190/AMF/1246028, https://doi.org/10.17190/ AMF/1480317, https://doi.org/10.17190/AMF/1419513, https://doi. org/10.17190/AMF/1246040, https://doi.org/10.17190/AMF/1246031, https://doi.org/10.17190/AMF/1246032, https://doi.org/10.17190/ AMF/1246036, https://doi.org/10.17190/AMF/1246038, https://doi. org/10.17190/AMF/1246039, https://doi.org/10.17190/AMF/1246043, https://doi.org/10.17190/AMF/1660339, https://doi.org/10.17190/ AMF/1246156, https://doi.org/10.17190/AMF/1246117, https://doi. org/10.17190/AMF/1419512, https://doi.org/10.17190/AMF/1246045, https://doi.org/10.17190/AMF/1246046, https://doi.org/10.17190/ AMF/1246047, https://doi.org/10.17190/AMF/1246119, https://doi. org/10.17190/AMF/1246050, https://doi.org/10.17190/AMF/1246053, https://doi.org/10.17190/AMF/1246054, https://doi.org/10.17190/ AMF/1246051, https://doi.org/10.17190/AMF/1246052, https://doi. org/10.17190/AMF/1246056, https://doi.org/10.17190/AMF/1246057, https://doi.org/10.17190/AMF/1246058, https://doi.org/10.17190/ AMF/1562389, https://doi.org/10.17190/AMF/1543379, https://doi.org/10.17190/AMF/1246065, https://doi.org/10.17190/ AMF/1246066, https://doi.org/10.17190/AMF/1617696, https://doi. org/10.17190/AMF/1498745, https://doi.org/10.17190/AMF/1634882,
https://doi.org/10.17190/AMF/1246070, https://doi.org/10.17190/ AMF/1660346, https://doi.org/10.17190/AMF/1246074, https://doi. org/10.17190/AMF/1246076, https://doi.org/10.17190/AMF/1246079, https://doi.org/10.17190/AMF/1246128, https://doi.org/10.17190/ AMF/1617715, https://doi.org/10.17190/AMF/1617716, https://doi. org/10.17190/AMF/1246080, https://doi.org/10.17190/AMF/1246081, https://doi.org/10.17190/AMF/1246083, https://doi.org/10.17190/ AMF/1419506, https://doi.org/10.17190/AMF/1480314, https://doi. org/10.17190/AMF/1246084, https://doi.org/10.17190/AMF/1246085, https://doi.org/10.17190/AMF/1246086, https://doi.org/10.17190/ AMF/1246088, https://doi.org/10.17190/AMF/1246089, https://doi. org/10.17190/AMF/1246092, https://doi.org/10.17190/AMF/1418683, https://doi.org/10.17190/AMF/1246093, https://doi.org/10.17190/ AMF/1419507, https://doi.org/10.17190/AMF/1419508, https://doi. org/10.17190/AMF/1419509, https://doi.org/10.17190/AMF/1617721, https://doi.org/10.17190/AMF/1617724, https://doi.org/10.17190/ AMF/1375201, https://doi.org/10.17190/AMF/1419502, https://doi. org/10.17190/AMF/1419501, https://doi.org/10.17190/AMF/1419504, https://doi.org/10.17190/AMF/1246136, https://doi.org/10.17190/ AMF/1246105, https://doi.org/10.17190/AMF/1246096, https://doi. org/10.17190/AMF/1418684, https://doi.org/10.17190/AMF/1246097, https://doi.org/10.17190/AMF/1246098, https://doi.org/10.17190/ AMF/1246099, https://doi.org/10.17190/AMF/1246101, https://doi. org/10.17190/AMF/1246102, https://doi.org/10.17190/AMF/1246127, https://doi.org/10.17190/AMF/1246154, https://doi.org/10.17190/ AMF/1246104, https://doi.org/10.17190/AMF/1418685, https://doi. org/10.17190/AMF/1660351, https://doi.org/10.17190/AMF/1246148, https://doi.org/10.17190/AMF/1246149, https://doi.org/10.17190/ AMF/1246140, https://doi.org/10.17190/AMF/1245984, https://doi. org/10.17190/AMF/1246109, https://doi.org/10.17190/AMF/1246111, https://doi.org/10.17190/AMF/1246112, https://doi.org/10.17190/ AMF/1617728, https://doi.org/10.17190/AMF/1579721, https://doi. org/10.17190/AMF/1617732, https://doi.org/10.17190/AMF/1579723, https://doi.org/10.17190/AMF/1617735, https://doi.org/10.17190/ AMF/1617737, https://doi.org/10.17190/AMF/1617741, https://doi. org/10.3133/sir20095079, https://doi.org/10.3133/sir20095079, https://doi.org/10.3133/sir20055288, https://doi.org/10.3133/ sir20055288, https://doi.org/10.3133/sir20085116, https://doi. org/10.3133/sir20095079, https://doi.org/10.3133/sir20085116, https:// doi.org/10.5066/F7R49NZN, https://doi.org/10.5066/F7R49NZN, https://doi.org/10.5066/F79C6WM9, https://doi.org/10.5066/ F79C6WM9, https://doi.org/10.3133/pp1805, https://doi.org/10.5066/ P9NZ9XSP, https://doi.org/10.5066/P9NZ9XSP, https://doi. org/10.5066/P9NZ9XSP, https://doi.org/10.3133/sir20075078, https:// doi.org/10.3133/sir20075078, https://doi.org/10.3133/sir20085116, https://doi.org/10.3133/sir20075078 and https://doi.org/10.3133/ sir20075078. Funding for AmeriFlux data resources was provided by the US Department of Energy Office of Science. Any use of trade, firm or product names is for descriptive purposes only and does not imply endorsement by the US Government.

Author contributions

F.S.M., J.L.H., J.M.V., R.A., M.A., J.B.F., A.K., A.R., G.B.S. and C.P. designed and guided the study; J.M.V., F.S.M., M.A. and L.J. wrote the main text; J.M.V. performed statistical analyses; C.M., J.M.V., B.M., T.O., C.D. and T.W. prepared measured data or model input data, or ran models; F.S.M., R.A., M.A., J.B.F., A.K., A.R., G.B.S., J.L.H., C.M., W.C., C.T.D., M.F., A.G., C.H., G.H., L.J., Y.K., K.K., S.O.-S., G.E.L.P., A.P., P.R., Y.Y., L.L. and B.C.d.A. developed models and OpenET infrastructure; J.M.V., M.A., F.S.M., L.J., R.A., J.B.F., J.L.H., A.K., G.B.S., T.O., B.M., A.R., M.F. and T.W. reviewed and edited text and figures.

Competing interests

The authors declare no competing interests.

Additional information

Extended data is available for this paper at
https://doi.org/10.1038/s44221-023-00181-7.
Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s44221-023-00181-7.
Correspondence and requests for materials should be addressed to John M. Volk.
Peer review information Nature Water thanks Tilden Meyers, Dennis Baldocchi and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
This is a U.S. Government work and not under copyright protection in the US; foreign copyright protection may apply 2024
Desert Research Institute, Reno, NV, USA. NASA Ames Research Center, Moffett Field, CA, USA. California State University Monterey Bay, Seaside, CA, USA. University of Idaho, Kimberly, ID, USA. USDA Agricultural Research Service, Beltsville, MD, USA. University of California, Los Angeles, Los Angeles, CA, USA. University of Nebraska-Lincoln, Lincoln, NE, USA. Universidade Federal do Rio Grande do Sul, Porto Alegre, RS, Brazil. US Geological Survey Earth Resources Observation and Science Center, North Central Climate Adaptation Science Center, Fort Collins, CO, USA. KBR, Inc. under contract to the US Geological Survey Earth Resources Observation and Science Center, Sioux Falls, SD, USA. NASA Marshall Space Flight Center, Huntsville, AL, USA. NASA Jet Propulsion Lab, Pasadena, CA, USA. University of California Berkeley, Berkeley, USA. USDA Agricultural Research Service, Davis, CA, USA. Innovate!, Inc. under contract to the US Geological Survey Earth Resources Observation and Science Center, Sioux Falls, SD, USA. Mississippi State University, Starkville, MS, USA. e-mail: john.volk@dri.edu

Evergreen Forests

Extended Data Fig. 1 | Monthly climatology of paired modeled and observed ET for evergreen forest sites. Subplot (a) shows monthly climatology of paired OpenET and flux tower from evergreen forested sites. Subplot (b) shows the residual of monthly mean ET (model minus mean closed flux ET). Unclosed
and closed labels refer to flux tower ET before and after energy balance closure correction. Dashed lines represent the closed flux ET mean plus two standard errors of the mean and unclosed flux ET mean minus two standard errors of the mean.

Mixed Forests

Extended Data Fig. 2 | Monthly climatology of paired modeled and observed ET for mixed forest sites. Subplot (a) shows monthly climatology of paired OpenET and flux tower from mixed forested sites. Subplot (b) shows the residual of monthly mean ET (model minus mean closed flux ET). Unclosed
and closed labels refer to flux tower ET before and after energy balance closure correction. Dashed lines represent the closed flux ET mean plus two standard errors of the mean and unclosed flux ET mean minus two standard errors of the mean.

Grasslands

Extended Data Fig. 3 | Monthly climatology of paired modeled and observed
ET for grassland sites. Subplot (a) shows monthly climatology of paired OpenET and flux tower ET from grassland sites. Subplot (b) shows the residual of monthly mean ET (model minus mean closed flux ET). Unclosed
and closed labels refer to flux tower ET before and after energy balance closure correction. Dashed lines represent the closed flux ET mean plus two standard errors of the mean and unclosed flux ET mean minus two standard errors of the mean.

Shrublands

Extended Data Fig. 4 | Monthly climatology of paired modeled and observed
ET for shrubland sites. Subplot (a) shows monthly climatology of paired OpenET and flux tower from shrubland sites. Subplot (b) shows the residual of monthly mean ET (model minus mean closed flux ET). Unclosed
and closed labels refer to flux tower ET before and after energy balance closure correction. Dashed lines represent the closed flux ET mean plus two standard errors of the mean and unclosed flux ET mean minus two standard errors of the mean.

Wetland/Riparian

Ensemble
DisALEXI eeMETRIC
geeSEBAL
PT-JPL
SSEBop
Closed + 2SE
Closed
Unclosed
– – Unclosed – 2SE

Month

Extended Data Fig. 5 | Monthly climatology of paired modeled and observed ET for wetland and riparian sites. Subplot (a) shows monthly climatology of paired OpenET and flux tower from wetland and riparian sites. Subplot (b) shows the residual of monthly mean ET (model minus mean closed flux ET).
Unclosed and closed labels refer to flux tower ET before and after energy balance closure correction. Dashed lines represent the closed flux ET mean plus two standard errors of the mean and unclosed flux ET mean minus two standard errors of the mean.
Extended Data Fig. 6 | Monthly climatology of modeled ET using all cropland pixels. Monthly climatology of OpenET ensemble members and the ensemble mean using all monthly ET data for all pixels that were classified as croplands for each year from 2016-2022.
Average Model Count in Ensemble
Extended Data Fig. 7| See next page for caption.
Extended Data Fig. 7| Spatial analysis of model ensemble outlier occurrence in cropland pixels. Subplot (a) shows the spatial differences between the OpenET ensemble mean growing season (April through October) ET for cropland pixels using the median absolute deviation (MAD) outlier removal approach and the simple arithmetic mean (SAM); monthly ET from 2016-2022
was used to build the map. Subplot (b) shows the average count of models used in the ensemble after outlier removal using all growing season monthly data for cropland pixels. A value of six indicates that no model was identified as an outlier, while four is the lower limit where a maximum of two models were removed as outliers before taking the ensemble mean.
Extended Data Fig. Spatial difference between mean growing season ET for each model from the ensemble value in cropland pixels. Difference between mean growing season (April through October) ET from each OpenET
model minus the ensemble mean using all monthly data from all pixels that were classified as croplands for each year from 2016-2022. See Supplementary Discussion 4 for a discussion of the Landsat striping exhibited by geeSEBAL.

natureportfolio

Corresponding author(s):
John Volk
Last updated by author(s): Nov 27, 2023

Reporting Summary

Nature Portfolio wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency in reporting. For further information on Nature Portfolio policies, see our Editorial Policies and the Editorial Policy Checklist.

Statistics

For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section.

Confirmed

The exact sample size ( ) for each experimental group/condition, given as a discrete number and unit of measurement
A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly
The statistical test(s) used AND whether they are one- or two-sided
Only common tests should be described solely by name; describe more complex techniques in the Methods section.
A description of all covariates tested
A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons
A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals)
For null hypothesis testing, the test statistic (e.g. F, ) with confidence intervals, effect sizes, degrees of freedom and value noted Give values as exact values whenever suitable.
For Bayesian analysis, information on the choice of priors and Markov chain Monte Carlo settings
For hierarchical and complex designs, identification of the appropriate level for tests and full reporting of outcomes
Estimates of effect sizes (e.g. Cohen’s , Pearson’s ), indicating how they were calculated
Our web collection on statistics for biologists contains articles on many of the points above.

Software and code

Policy information about availability of computer code

Data collection
The in situ measured evapotranspiration data analysed during the current study were processed using the “flux-data-qaqc” Python package, version 0.1.6 (https://github.com/Open-ET/flux-data-qaqc).
Data analysis
The “flux-data-footprint” Python package was used to generate temporally dynamic flux footprints for sampling of of daily and monthly model ET data (https://github.com/Open-ET/flux-data-footprint). The Numpy (version 1.17.2) and statsmodels (version 0.12.1) Python packages were used for data analyses during the current study.
For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors and reviewers. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Portfolio guidelines for submitting code & software for further information.

Data

Policy information about availability of data
All manuscripts must include a data availability statement. This statement should provide the following information, where applicable:
  • Accession codes, unique identifiers, or web links for publicly available datasets
  • A description of any restrictions on data availability
  • For clinical datasets or third party data, please ensure that the statement adheres to our policy
zenodo.7636781. The OpenET model ET data analyzed during the current study are available in the Zenodo repository, with identifier http://dx.doi.org/10.5281/ zenodo. 10119477.

Human research participants

Policy information about studies involving human research participants and Sex and Gender in Research.
Reporting on sex and gender The current study did not involve human participants, their data, or their biological material.
Population characteristics The current study did not involve human participants, their data, or their biological material.
Recruitment The current study did not involve human participants, their data, or their biological material.
Ethics oversight The current study did not involve human participants, their data, or their biological material.
Note that full information on the approval of the study protocol must also be provided in the manuscript.

Field-specific reporting

Please select the one below that is the best fit for your research. If you are not sure, read the appropriate sections before making your selection.
Life sciences Behavioural & social sciences
Ecological, evolutionary & environmental sciences
For a reference copy of the document with all sections, see nature.com/documents/nr-reporting-summary-flat.pdf

Ecological, evolutionary & environmental sciences study design

All studies must disclose on these points even when the disclosure is negative.
Study description The current study is focused on one-to-one comparisons between evapotranspiration data as modeled by remote-sensing methods and as measured on the ground; well known goodness-of-fit metrics were used to evaluate model data against measured data. Modeled and measured data were paired based on temporally overlapping records at multiple measurement stations. Accuracy metric results were grouped by different shared characteristics of the stations, using weighted averaging. Spatial long-term model data were mapped for individual models and differenced from the model ensemble average.
Research sample The measured daily and monthly evapotranspiration data in the current study came from a public dataset that is available here: http://dx.doi.org/10.5281/zenodo.7636781. The measurements used were collected between 1995-2021. OpenET model data were generated and paired to the daily and monthly measurements, however model data was not available prior to 2001. The total number of stations with paired data was 152, and among them there were 16,444 days and 4,107 months of paired data.
Sampling strategy As a conservative measure, we required a minimum of 3 months of paired data per measurement station to be included in weighted average accuracy metrics. In order to avoid skewing grouped average metrics, we weighted each station by the square root of the number of paired data.
Data collection Measured ET data was previously curated and publicly archived on Zenodo (http://dx.doi.org/10.5281/zenodo.7636781). Model ET data was generated for the study using the current OpenET version.
Timing and spatial scale Measured data was collected primarily from eddy covariance systems and post-processed to daily and monthly aggregated periods. Model ET was sampled at daily (the dates of satellite overpass) and monthly intervals at the measurement stations using pixel footprints which were determined by either long-term wind direction or from a physically based flux footprint prediction model. Flux footprints rarely exceeded the size of a ( 30 m resolution) grid. Gaps in measured data exist due to lapses in sensor operation or faulty data, and gaps exist in model data due to cloud coverage.
Data exclusions No data available to us were excluded from the analyses during the current study.
Reproducibility Data processing steps for both modeled and measured data used in the current study were made reproducible by our use of well-documented open source software. We developed Python code for eddy covariance data processing and footprint development. Similarly, the OpenET modeled data was generated using the operational models which are open source and their data are also publicly available through various mechanisms including the OpenET API and the Google Earth Engine Data Catalog. The statistical methods used in the study were limited to simple techniques, and the results were independently tested by multiple members of the OpenET group using different statistical packages such as Python and Microsoft Excel.
Randomization Randomization of data into groups was not applicable as groups were defined by biophysical characteristics such as climate and land cover type.
Blinding This study did not employ blinding of data due to the limited amount of high quality measured evapotranspiration data available to us. However, additional data that has not yet been compared against OpenET models was held out of this study for a future blind intercomparison and accuracy assessment of OpenET.

Reporting for specific materials, systems and methods

We require information from authors about some types of materials, experimental systems and methods used in many studies. Here, indicate whether each material, system or method listed is relevant to your study. If you are not sure if a list item applies to your research, read the appropriate section before selecting a response.
Materials & experimental systems Methods
n/a Involved in the study n/a Involved in the study
Antibodies
ChIP-seq
Palaeontology and archaeology

  1. The in situ measured evapotranspiration data analysed during the current study are available in the Zenodo repository, with identifier http://dx.doi.org/10.5281/