Evaluating ECOSTRESS data across the Australia and Korea

The effects of extreme weather events due to climate change are causing localized energy imbalances (that are) affecting evapotranspiration and drought. Thus, quantifying hydrological cycle components is essential for efficient water resource management. Generally, hydrometeorological variables are acquired from point-based observations, while it has limitations in representing the spatial distribution of hydrometeorological variables. As an alternative, remote sensing imagery has been widely utilized to overcome the limitation. Remote sensing-based land surface temperature (LST) and evapotranspiration (ET) have been estimated by the Moderate-resolution Imaging Spectro-radiometer (MODIS) sensor operated by the National Aeronautics and Space Administration (NASA) since 1999. However, the MODIS sensor's coarse spatial resolution (LST: 500 m, 1 km; ET: 500 m) limits its ability to capture the spatial distribution of hydrometeorological variables over complex terrain. On the other hand, Ecosystem Spaceborne Thermal Radiometer Experiment on Space Station (ECOSTRESS) developed by NASA Jet Propulsion Laboratory and launched in 2018, provides a variety of outputs (LST, ET, etc.) at a higher spatial resolution (70m) than existing MODIS outputs. The main purpose of this study is to evaluate the applicability of ECOSTRESS LST and ET by comparing against eddy covariance-based flux tower observations (from 25 stations) as well as MODIS products across Korea and Australia from June 2018 to December 2022. The comparison of ECOSTRESS LST against flux tower LST revealed similar trends in Korea (Correlation coefficient R: 0.64, Index of Agreement IOA: 0.77) compared to Australia (R: 0.26, IOA: 0.32). In terms of magnitude, ECOSTRESS LST showed underestimation with high root mean square error (RMSE) for both Australia (bias: -8.05, RMSE: 19.22) and Korea (bias: -4.19, RMSE: 10.73). Seasonal behavior of ECORSTRESS LST showed the highest uncertainty during summer for both Australia and Korea. For the Australia, either forest or grassland sites located in northern part (classified as tropical or arid climate zone) of Australia revealed high magnitude of bias and RMSE. Evaluation of ECOSTRESS daily ET by comparing to latent heat (LE) measured from flux towers yielded a poor agreement over both Australia (bias: 4.92 mm/day, RMSE: 6.59 mm/day, R: 0.14, IOA: 0.17) and Korea (bias: 8.80 mm/day, RMSE: 11. 61 mm/day, R: -0.02, IOA: 0.12) with the positive bias indicating that the ECOSTRESS ET is overestimated. Spatial analysis of error statistics revealed that northern tropical area over Australia with high precipitation during summer yielded high magnitude of bias and RMSE. Overall result showed that ECOSTRESS LST and ET tended to be underestimated and overestimated, respectively. For the Australia, northern part of Australia classified as tropical zone yielded highest uncertainty for both ET and LST. Therefore, it is judged that additional validation and calibration processes with consideration of various geomorphological and hydrological characteristics should be performed to increase the applicability of ECOSTRESS outputs. Acknowledgement: This research was supported by Korea National University of Transportation in 2024.