2020 ESA Annual Meeting (August 3 - 6)

PS 48 Abstract - Integration of state-of-the-art approaches in remote sensing, machine learning, and land surface models to evaluate global terrestrial evapotranspiration

Shufen Pan1, Naiqing Pan2, Hanqin Tian1, Pierre Friedlingstein3, Stephen Sitch4, Hao Shi2, Vivek Arora5, Vanessa Haverd6, Atul Jain7, Etsushi Kato8, Sebastian Lienert9, Danica Lombardozzi10, Julia Nabel11, Catherine Ottlé12, Benjamin Poulter13, Sönke Zaehle14 and Steve W. Running15, (1)International Center for Climate and Global Change Research and School of Forestry and Wildlife Sciences, Auburn University, Auburn, AL, (2)Auburn University, Auburn, AL, (3)University of Exeter, (4)University of Leeds, (5)Canadian Centre for Climate Modelling and Analysis, (6)CSIRO Oceans and Atmosphere, (7)Department of Atmospheric Sciences, University of Illinois, Urbana, IL, (8)Institute of Applied Energy (IAE), (9)Climate and Environmental Physics, Physics Institute, University of Bern, (10)CGD, National Center for Atmospheric Research, Boulder, CO, (11)Max Planck Institute for Meteorology, (12)LSCE-IPSL-CNRS, Orme des Merisiers, (13)Biospheric Sciences Laboratory, NASA Goddard Space Flight Center, Greenbelt, MD, (14)Department of Biogeochemical Systems, Max-Planck Institute for Biogeochemistry, Jena, Germany, (15)Ecosystem and Conservation Sciences, University of Montana, Missoula, MT
Background/Question/Methods

Evapotranspiration (ET) is a critical component in the global water cycle and links terrestrial water, carbon and energy cycles. Accurate estimate of terrestrial ET is important for hydrological, meteorological, and agricultural research and applications. However, direct measurement of global terrestrial ET is not feasible. In this study, we integrate state-of-the-art estimates of global terrestrial ET, including data-driven and process-based estimates: 1) remote sensing-based physical models, 2) machine learning methods, and 3) land surface models (LSMs), to assess its spatial pattern, inter-annual variability, environmental drivers, long-term trend, and response to vegetation greening. We evaluated twenty global terrestrial ET estimates including four from remote sensing-based physical models, two from machine-learning algorithms and fourteen from TRENDY LSMs. Our goal is not to compare the various models and choose the best one, but to identify the uncertainty sources in each type of estimate and provide suggestions for future model development.

Results/Conclusions

The results showed that the ensemble means of annual global terrestrial ET estimated by these three categories of approaches agreed well, ranging from 589.6 mm yr-1 to 617.1 mm yr-1. LSMs had much uncertainty in the ET magnitude in tropical regions especially the Amazon Basin, while remote sensing-based ET products showed larger inter-model range in arid and semi-arid regions than LSMs. LSMs and remote sensing-based physical models exhibited much larger inter-annual variability (IAV) in ET than machine learning based algorithms in southwestern U.S. and the Southern Hemisphere, particularly in Australia. All the twenty models used in this study showed anthropogenic earth greening had a positive role in increasing terrestrial ET. The concurrent small inter-annual variability, i.e. relative stability, found in all estimates of global terrestrial ET, suggests there exists a potential planetary boundary in regulating global terrestrial ET, with the value being about 6.74×104 km3 yr-1 (603mm yr-1). Uncertainties among approaches were identified in specific regions, particularly in the Amazon Basin and arid/semi-arid regions. Improvements in parameterizing water stress and canopy dynamics, utilization of new available satellite retrievals and deep learning methods, and model-data fusion will advance our predictive understanding of terrestrial ET estimates.