Satellite data-driven modeling of field scale evapotranspiration in croplands using the MOD16 algorithm framework

Abstract. Evapotranspiration (ET) is a crucial component of the terrestrial hydrological cycle. Latent heat flux (LE, equivalent to ET in W m−2) observed by the eddy covariance (EC) technique, commonly known as LEEC, has been widely recognized as a highly accurate benchmark for global ET estimation. Currently, there is an increasing need for long-time-series benchmark data to support climate change analysis, construction of new models, and validation of new products. However, existing LEEC datasets, like FLUXNET2015, face significant challenges due to limited observation periods and extensive data gaps, which hinders their application in ET modeling and global change analysis. To address these issues, we developed a gap-filling and prolongation framework for LEEC data and established a benchmark dataset for global ET estimation from 2000 to 2022 across 64 sites at various timescales. The framework mainly includes three parts: site selection and data pre-processing, generation of gap-filled half-hourly/hourly LE data, and generation of prolonged daily LE data. We selected 64 sites from FLUXNET2015 based on rigorous filtering criteria. A novel bias-corrected random forest (RF) algorithm was used for gap-filling and prolongation in the framework to produce seamless half-hourly and daily LE data. After analysis, the framework using the novel bias-corrected RF algorithm achieves excellent performance in both hourly gap-filling and daily prolongation, with mean root mean square error values of 33.86 and 16.58 W m−2, respectively. The algorithm significantly improves the gap-filling performance for long gaps and extreme values compared with the original RF and marginal distribution sampling algorithm. The results demonstrate robust prolongation performance of our framework in both prolongation directions and temporal stability. Furthermore, our gap-filled dataset demonstrates strong consistency with FLUXNET2015 in terms of data distribution. In conclusion, we have published the first benchmark dataset for global ET estimation based on FLUXNET2015 from 2000 to 2022. This dataset can effectively provide data support for ET modeling, water–carbon cycle monitoring, and climate change analysis. It is made freely available via the following repository: https://doi.org/10.5281/zenodo.13853409 (Li et al., 2024b).

Accurate quantification of evapotranspiration (ET) is crucial for understanding variability in the global water cycle, yet state-of-the-art estimates of ET derived from models and remote sensing products contain large uncertainties. Taking the advantage of extensive eddy covariance measurements and machine learning algorithms, ET can be upscaled from globally distributed in-situ observations by combining them with global meteorological and satellite data (e.g., FLUXCOM ensembles, Jung et al., 2019). However, eddy covariance measurements suffer from well-known energy balance non-closure problems, and those uncertainties are further propagated to the global ET estimates. Here, we first estimate the energy balance non-closure within dynamic sliding windows for flux tower site, then we compute correction factors for ET measurements following different hypothesis (that assign errors to latent and/or sensible heat fluxes) according to insights from large eddy simulation studies. Then energy balance closure corrected ET data are used in FLUXCOM to estimate global ET. The upscaled ET then is evaluated by comparison with water-balance-drived ET at the catchment level. This comparison helps to determine the most consistent correction of ET for different regions and conditions. By providing improved global ET estimates, water-related studies can be further facilitated, and model parameterizations can be further optimized to address the challenges posed by climate change on ecosystems and water resources.

Global estimates of evapotranspiration and gross primary production based on MODIS and global meteorology data

Abstract. Accurate global gridded estimates of evapotranspiration (ET) are key to understanding water and energy budgets, in addition to being required for model evaluation. Several gridded ET products have already been developed which differ in their data requirements, the approaches used to derive them and their estimates, yet it is not clear which provides the most reliable estimates. This paper presents a new global ET dataset and associated uncertainty with monthly temporal resolution for 2000–2009. Six existing gridded ET products are combined using a weighting approach trained by observational datasets from 159 FLUXNET sites. The weighting method is based on a technique that provides an analytically optimal linear combination of ET products compared to site data and accounts for both the performance differences and error covariance between the participating ET products. We examine the performance of the weighting approach in several in-sample and out-of-sample tests that confirm that point-based estimates of flux towers provide information on the grid scale of these products. We also provide evidence that the weighted product performs better than its six constituent ET product members in four common metrics. Uncertainty in the ET estimate is derived by rescaling the spread of participating ET products so that their spread reflects the ability of the weighted mean estimate to match flux tower data. While issues in observational data and any common biases in participating ET datasets are limitations to the success of this approach, future datasets can easily be incorporated and enhance the derived product.

Evapotranspiration (ET) plays a significant role in monitoring and forecasting drought. High spatiotemporal resolution ET data enables precise decision-making in agricultural management and drought assessment. The synergy of daily Sentinel-3 Land Surface Temperature (LST) and Sentinel-2 multispectral images offers opportunities for more accurate ET estimation. In this study, we estimate and assess ET of Busia County, Kenya using Earth observation data, machine learning, and Two-Source Energy Balance (TSEB) model. A random forest model is employed with an input of Sentinel-2 spectral indices to downscale Sentinel-3 1-km LST images to 20-m. An optimized random forest model produced 20-m LST with accuracies 0.88, 0.82, and 0.55 for R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>, RMSE, and MAE. The finer resolution of LST and ERA5 meteorological data serve as inputs for TSEB, estimating daily actual ET at 20-m. The final ET product (20m, Sentinel 3 timesteps) is compared with WaPOR and Gleam products to evaluate the derived ET products.

A universal triangle method for evapotranspiration estimation with MODIS products and routine meteorological observations: Algorithm development and global validation

Abstract. Land evapotranspiration (ET) estimates are available from several global data sets. Here, monthly global land ET synthesis products, merged from these individual data sets over the time periods 1989–1995 (7 yr) and 1989–2005 (17 yr), are presented. The merged synthesis products over the shorter period are based on a total of 40 distinct data sets while those over the longer period are based on a total of 14 data sets. In the individual data sets, ET is derived from satellite and/or in situ observations (diagnostic data sets) or calculated via land-surface models (LSMs) driven with observations-based forcing or output from atmospheric reanalyses. Statistics for four merged synthesis products are provided, one including all data sets and three including only data sets from one category each (diagnostic, LSMs, and reanalyses). The multi-annual variations of ET in the merged synthesis products display realistic responses. They are also consistent with previous findings of a global increase in ET between 1989 and 1997 (0.13 mm yr−2 in our merged product) followed by a significant decrease in this trend (−0.18 mm yr−2), although these trends are relatively small compared to the uncertainty of absolute ET values. The global mean ET from the merged synthesis products (based on all data sets) is 493 mm yr−1 (1.35 mm d−1) for both the 1989–1995 and 1989–2005 products, which is relatively low compared to previously published estimates. We estimate global runoff (precipitation minus ET) to 263 mm yr−1 (34 406 km3 yr−1) for a total land area of 130 922 000 km2. Precipitation, being an important driving factor and input to most simulated ET data sets, presents uncertainties between single data sets as large as those in the ET estimates. In order to reduce uncertainties in current ET products, improving the accuracy of the input variables, especially precipitation, as well as the parameterizations of ET, are crucial.

Evapotranspiration (ET) is critical to surface water dynamics. Effective water resource management necessitates an accurate ET estimation. In the Yellow River Basin China, a study area, cutting-edge technologies are needed to improve large-scale ET estimates. This study estimates ET using GSEBAL, an advanced ET estimation algorithm. Google Earth Engine integrates the surface energy balance model-based GSEBAL. The technique includes the collection, preparation, and calculation of ET using Landsat imagery and ERA5-Land meteorological data from 1990 to 2020. The study examined satellite LST, albedo, and NDVI data. The GSEBAL model calculates soil heat flow, net radiation, and sensible heat flux. The study tested the GSEBAL model utilizing essential ET datasets such as ECOSTRESS, MOD16, and SSEBop. The study showed that the model effectively predicted daily and seasonal ET variations in different climates. Root mean squared error, bias, and Pearson's correlation coefficient verified the model's reliability. The study also analyzed land use and land cover (LULC) over 30 years using Random Forest classifiers. In the 1990–2020 YRBC ET, land use changes affect ET rates annually and seasonally. The study area experiences changes in LST, NDVI, and LULC. Maximum ET values rose from 214.217 mm in 1990 to 234.891 mm in 2000. The pattern flipped in 2020, decreasing to 221.456 mm. In 2010, Summer had the highest ET, 484.455 mm. 2020 spring ET is 314.727 mm. Low ET decreased from 24.652 mm in 1990 to 18.2 mm in 2020, reducing water loss. Fall ET peaks at 24.9 mm in 2020; winter ET is 18.75 mm.

Analysis of various surface energy balance models for evapotranspiration estimation using satellite data

Coupling physical constraints with machine learning for satellite-derived evapotranspiration of the Tibetan Plateau

Actual evapotranspiration (ET) is a major water use flux in a basin water balance with crucial significance for water resources management and planning. Mapping ET with good accuracy has been the subject of ongoing research. Such mapping is even more challenging in heterogeneous and data-scarce regions. The main objective of our research is to estimate ET using daily Moderate Resolution Imaging Spectroradiometer (MODIS) land surface temperature and Global Land Data Assimilation System (GLDAS) weather datasets based on the operational simplified surface energy balance (SSEBop) algorithm at a 1-km spatial scale and 8-day temporal resolution for the Mara Basin (Kenya/Tanzania). Unlike previous studies where the SSEBop algorithm was used, we use a seasonally-varying calibration coefficient for determining the “cold” reference temperature. Our results show that ET is highly variable, with a high inter-quartile range for wetlands and evergreen forest (24% to 29% of the median) and even up to 52% of the median for herbaceous land cover and rainfed agriculture. The basin average ET accounts for about 66% of the rainfall with minimal inter-annual variability. The basin scale validation using nine-years of monthly, gridded global flux tower-based ET (GFET) data reveals that our ET is able to explain 64% of the variance in GFET while the MOD16-NB (Nile Basin) explains 72%. We also observe a percent of bias (PBIAS) of 1.1% and 2.8%, respectively for SSEBop ET and MOD16-NB, indicating a good reliability in the ET estimates. Additionally, the SSEBop ET explains about 52% of the observed variability in the Normalized Difference Vegetation Index (NDVI) for a 16-day temporal resolution and 81% for the annual resolution, pointing to an increased reliability for longer aggregation periods. The annual SSEBop ET estimates are also consistent with the underlying primary (i.e., water and energy) and secondary (i.e., soil, topography, geology, land cover, etc.) controlling factors across the basin. This paper demonstrated how to effectively estimate and evaluate spatially-distributed and temporally-varying ET in data-scarce regions that can be applied elsewhere in the world where observed hydro-meteorological variables are limited.

An accurate estimation of evapotranspiration (ET) is critical to understanding the water cycle in watersheds and promoting the sustainable utilization of water resources. Although there are various ET products in the Yellow River Basin, various ET products have many uncertainties due to input data, parameterization schemes, and scale conversion, resulting in significant uncertainties in regional ET data products. To reduce the uncertainty of a single product and obtain more accurate ET data, more accurate ET data can be obtained by fusing different ET data. Addressing this challenge, by calculating the uncertainty of three ET data products, namely global land surface satellite (GLASS) ET, Penman–Monteith–Leuning (PML)-V2 ET, and reliability-affordable averaging (REA) ET, the weight of each product is obtained to drive the Bayesian three-cornered Hat (BTCH) algorithm to obtain higher quality fused ET data, which are then validated at the site and basin scales, and the accuracy has significantly improved compared to a single input product. On a daily scale, the fused data’s root mean square error (RMSE) is 0.78 mm/day and 1.14 mm/day. The mean absolute error (MAE) is 0.53 mm/day and 0.84 mm/day, respectively, which has a lower RMSE and MAE than the model input data; the correlation coefficients (R) are 0.9 and 0.83, respectively. At the basin scale, the RMSE and MAE of the annual average ET of the fused data are 11.77 mm/year and 14.95 mm/year, respectively, and the correlation coefficient is 0.84. The results show that the BTCH ET fusion data are better than single-input product data. An analysis of the fused ET data on a spatiotemporal scale shows that from 2001 to 2017, the ET increased in 85.64% of the area of the Yellow River Basin. Fluctuations in ET were greater in the middle reaches of the Yellow River than in the upstream and downstream regions. The BTCH algorithm has indispensable reference value for regional ET estimation research, and the ET data after BTCH algorithm fusion have higher data quality than the original input data. The fused ET data can inform the development of management strategies for water resources in the YRB and provide a deeper understanding of the regional water supply and demand balance mechanism.

An ensemble data assimilation approach to improve farm-scale actual evapotranspiration estimation

Agriculture on the Texas High Plains (THP) uses approximately 89% of groundwater withdrawals from the Ogallala Aquifer, leading to steady decline in water table levels. Therefore, efficient water management is essential for sustaining agricultural production in the THP. Accurate evapotranspiration (ET) maps provide critical information on actual spatio‐temporal crop water use. METRIC (Mapping Evapotranspiration at High Resolution using Internalized Calibration) is a remote sensing based energy balance method that uses radiometric surface temperature (T s) for mapping ET. However, T s calibration effects on satellite based ET estimation are less known. Further, METRIC has never been applied for the advective conditions of the semi‐arid THP. In this study, METRIC was applied and predicted ET was compared with measured values from five monolithic weighing lysimeters at the USDA‐ARS Conservation and Production Research Laboratory in Bushland, Texas, USA. Three different levels of calibration were applied on a Landsat 5 Thematic Mapper's thermal image acquired on 23 July 2006 to derive T s. Application of METRIC on a MODTRAN calibrated image improved the accuracy of distributed ET prediction. In addition, ET estimates were further improved when a THP‐specific model was used for estimating leaf area index. Results indicated that METRIC performed well with ET mean bias error±root mean square error of 0.4±0.7 mm d−1.

Due to limited direct measurements, regional or global terrestrial evapotranspiration (ET) is generally derived from a combination of meteorological and satellite observations. Although the inhomogeneity of the observed climate data has been widely reported, its impact on the calculated ET has not been adequately quantified. This study aimed to calculate ET using the modified Penman‐Monteith (MPM) model with raw and homogenized meteorological data. Additionally, we compared the calculated ET with those estimates from variable methods (water balance, satellite‐based, and reanalysis) in China and its six major river basins from 1982 to 2020. During the overlapping period of 1997–2018, ET calculated from raw input data decreased slightly at −0.39 mm yr−2 (p = 0.64) in China, whereas homogenized ET showed a significant increasing trend of 0.93 mm yr−2 (p = 0.02), with a better agreement with water balance ET (1.93 mm yr−2, p = 0). Global Land Evaporation Amsterdam Model (GLEAM) and Modern‐Era Retrospective Analysis for Research and Applications, version 2 (MERRA2) could reproduce the increasing trends with 2.08 mm yr−2 (p = 0) and 2.59 mm yr−2 (p = 0). The intercomparison of input variables (solar radiation, relative humidity, wind speed, precipitation, and air temperature) among ET products revealed substantial differences, which can account for the discrepancies in ET estimates. Homogenized ET, GLEAM and MERRA2 exhibited significant increasing trends in China and most river basins from 1982 to 2020. Our findings underscore the importance of utilizing homogenized input data for more accurate ET estimation.