38HYDRO Landsat Imagery Shows a Recent Accelerating Rate of Increase in Land Evapotranspiration.

Evapotranspiration (ET) is the second largest component of the global water cycle¹ and it plays a vital role in Earth’s surface energy balance¹. It controls the availability and distribution of water on Earth’s surface, maintains water balance, and helps regulate the temperature of the Earth’s surface through its cooling effect. Spatially, ET is usually estimated using methods such as remote sensing (RS)², water budget³, and machine learning^4,5. Approaches to estimate ET bear many challenges and uncertainties due to many factors such as input data sources and modeling assumptions⁶. Studies have shown that global warming has largely contributed to the intensification of the water cycle by altering global ET⁷, precipitation patterns⁸, runoff⁹, atmospheric water vapor¹⁰, and melting of ice glaciers¹¹. The 10 warmest years on record have all occurred since 2010, and there is evidence that these years have witnessed intensified ET^12,13, fueling atmospheric moisture¹⁰. In the face of a changing climate, a comprehensive understanding of evapotranspiration (ET) at both global historical and field-specific scales, as well as a careful analysis of recent ET changes, is imperative to effectively manage our water resources and ensure adequate crop yields for global food security. By better understanding the intricate dynamics of ET, we can make more informed decisions that positively impact the health of our ecosystems and the well-being of human communities around the world.

Discrepancies in historical trends of ET are mainly attributed to the scarcity of long-term ET data across the globe¹⁴, as well as fluctuations in ET estimation methods (e.g., water budget analysis vs. land surface model simulations) and underlying input data. Coarse resolution long-term ET retrievals exhibit a recent upward global trend^15–19, generally attributed to rising air temperatures¹⁵. Changes in ET are mainly driven by vegetation cover dynamics^6,14,20, anthropogenic drivers²¹, atmospheric demand²², precipitation¹⁴, net radiation²³, and natural variability associated with El-Nino Southern Oscillations (ENSO)¹⁵ and prevailing La Nina events²⁴. While many studies agree that the ET trend is increasing globally, others have noted a decreasing global ET over the past few decades^5,25,26. Historical trends in ET remain uncertain.

The results of these numerous studies confirm the significance of conducting an updated long-term assessment of ET trends using high-resolution satellite missions. Global ET has been estimated from coarse spatial resolution satellite observations (e.g., GLEAM¹⁵). Higher resolution ET maps have been produced only for specific locations using Landsat satellite imagery using remote sensing (e.g., ALEXI/DisALEXI²⁷, METRIC²⁸, PT-JPL²⁹, SEBAL³⁰, SIMS³¹ and SSEBop³²). Land surface temperature (LST)-based models capture moisture signals through the effect of evaporative cooling on the LST and possess lower sensitivity to meteorological inputs ⁶. Nevertheless, there is a lack of a globally consistent field-scale mapping of evapotranspiration which is crucial for quantifying and monitoring land water use and for improving our understanding of ET dynamics as related to a human-induced climate change.

Here, we generated and analyzed a monthly global 100-m ET dataset for 1990–2021 over land. We utilized the HSEB model detailed and validated in Jaafar, et al. ^1,6, that assimilates thermal and optical data from Landsat thermal and optical satellite imagery, land cover data from the European Space Agency global dynamic Land cover Climate Change Initiative dataset (ESA CCI – LC 300m)³³, and weather data from the European Centre for Medium-Range Weather Forecasts (ECMWF) Reanalysis hourly data (ERA-5 Land)³⁴, to calculate global ET at the 30-m spatial resolution and 8-day time scale. We interpolated the ET to the monthly scale and exported them at 100-m. We then validated the product against monthly ET from 193 FLUX site towers (N = 4483). We examined the trends in global ET and the possible drivers of ET with a special focus on 2000–2021 (to avoid potential bias caused by the lack of sufficient imagery cover in some regions in the 1990s). The dataset can be accessed for free as an image collection on Google Earth Engine (GEE), providing a valuable resource for research and applications in various fields, including but not limited to, water resource management, climate change, and agriculture.