Newswise — Urbana, Ill. – When scientists look at the Earth's availability of water for ecosystem services, they're not just looking at precipitation. We also need to consider the movement of water from the earth's surface to the atmosphere, a process known as evapotranspiration (ET). ET includes evaporation from the soil and outdoor pools such as lakes, rivers, and ponds, and from plant leaves. The difference between precipitation and ET indicates the balance of water available for social needs such as agricultural and industrial production. However, measuring ET is difficult. A new study from the University of Illinois at Urbana-Champaign presents a computer model that uses artificial intelligence (AI) to predict ET based on remote sensing estimates.
“Ground-based ET estimates capture local fluxes of water transferred into the atmosphere, but at limited scale. In contrast, satellite data provide ET information on a global scale. Still, it is often incomplete due to clouds and sensor failures, and satellite cycles in some regions can take several days. We can predict missing data and account for the dynamics of land use and atmospheric movement. We conducted this study to generate daily continuous ET data for the purpose of generating ET data,” said first author Jeongho Han, a doctoral student in the Department of Agricultural and Biological Engineering (ABE). Part of the Illinois College of Agricultural, Consumer and Environmental Sciences and Granger Institute of Technology.
The researchers created a “Dynamic Land Cover Evapotranspiration Modeling Algorithm” (DyLEMa) based on a decision tree machine learning model. This algorithm aims to predict missing spatial and temporal ET data using a trained seasonal machine learning model. DyLEMa was assessed to the size of Illinois on a 30 x 30 meter grid daily for 20 years using data from NASA, the U.S. Geological Survey, and the National Oceanic and Atmospheric Administration.
“DyLEMa is much more detailed and complex than other models. It distinguishes between different land uses such as forest, urban, and agriculture, and different crops such as corn and soybeans. The model includes precipitation, temperature, These include humidity, solar radiation, vegetation stage, and soil properties. This allows us to accurately capture surface dynamics and predict ET based on multiple variables. It is particularly important for agricultural landscapes,” said co-author Jorge Guzmán, research assistant professor at ABE.
The researchers tested the model's accuracy by comparing the model's results to existing data. For long-term validation, they used his ground measurements from 2009 to 2016 at his four sites in Illinois. To test the spatial accuracy, we also created an artificial scenario that inserted synthetic clouds into a cloudless image, applied the algorithm, and compared the results with the original data. Overall, DyLEMa reduces the ET prediction uncertainty in the cumulative ET estimate from an average of +30% (overprediction) to approximately -7% (underprediction) compared to existing measurements, with improved accuracy. It shows a significant improvement.
This research is part of a larger project on soil erosion funded by the USDA National Institute of Food and Agriculture. Maria Chu, an associate professor at ABE, is the principal investigator on that project and co-author of the new paper.
“ET controls soil water content and vice versa, which affects surface processes such as runoff and erosion. Our next step is to estimate soil erosion more accurately. It’s about integrating the data into a distributed hydrological model,” Chu said.
“One of the challenges in land management practice is that people may not immediately see the benefits of introducing changes. But using this model, we can see that what we're doing now, for example, Years or even 20 years later, we can show long-term impacts far from the farm. This is data and computing power to engage communities and inform policy measures. It is the power to harness the power of,” Chu added.
The researchers worked with the National Center for Supercomputing Applications (NCSA) and the Illinois Campus Cluster Program (ICCP) to process the data and train the model. They plan to make the data, including his 20 years of work at Illinois, available to other researchers.
The paper “Dynamic land cover evapotranspiration model algorithm: DyLEMa” Computers and electronics in agriculture [doi.org/10.1016/j.compag.2024.108875]. This research was funded by the U.S. Department of Agriculture-National Institute of Food and Agriculture (NIFA) award number 2019-67019-29884, the ACCESS program funded by the National Science Foundation, and his UIUC Illinois Campus Cluster.
