Author(s): Akash Atnurkar; Meenu Ramadas
Linked Author(s): AKASH ATNURKAR
Keywords: Soil moisture seasonal variability microwave remote sensing SMAP AMSR-2
Abstract: Surface soil moisture (SSM) data at regional scale is beneficial for agricultural drought assessment, optimal water resources allocation, fine-tuning irrigation schedules, hydrological modeling, and environmental studies. For decades, passive/active microwave sensors onboard different Earth Observation (EO) satellites have been employed to estimate SSM at global scale. The Soil Moisture Active Passive (SMAP) level 3 (L3.0) daily SM retrievals at 9-km spatial resolution and Advanced Microwave Scanning Radiometer (AMSR-2) L3 daily soil moisture retrievals at 10-km spatial resolution have been popularly adopted in many regional studies. However, their validation needs to be performed from time-to-time using observed SSM data. This study therefore targets a performance evaluation of these two SSM products based on seasonal characteristics, coefficient of variation (CV), and popular evaluation metrics. Firstly, for two agricultural study areas (SA), the observed daily dataset during 2016-2017 (SA-I) and 2023-2024 (SA-II), respectively were analyzed. Then, the SMAP and AMSR-2 products over these areas were processed so that their performance evaluation could be carried out by comparing with the observed field data. The analysis explored the seasonal fluctuations in the study areas’ SSM. Further, we considered the spatial variability using the CV values, and this variability was not found to be captured well by both SMAP and AMSR-2 when compared to the field data. After applying popular evaluation metrics such as mean absolute error (MAE), correlation coefficient (R), root mean square error (RMSE) and unbiased RMSE (ubRMSE), the results revealed that SMAP and AMSR-2 effectively captured seasonal soil moisture dynamics in summer with R=0.65 and 0.67 respectively, and RMSE of 0.12 m3/m3 and 0.15 m3/m3 respectively. Whereas, AMSR-2 performed better at capturing seasonal soil moisture variation in monsoon and post-monsoon seasons with R values of 0.62 and 0.51, respectively. Our findings provide valuable insights on the variability in remote sensing-based SSM products to aid comprehensive evaluation of their performance as proxies of observed data over agricultural regions.
Year: 2025