AI in soil moisture remote sensing Carsten Montzka, Luca Brocca, Hao Chen, Narendra N. Das, Antara Dasgupta, et al. International Journal of Applied Earth Observation and Geoinformation, 2026
Assessing evapotranspiration dynamics across central Europe in the context of land-atmosphere drivers Anke Fluhrer, Martin J. Baur, María Piles, Bagher Bayat, Mehdi Rahmati, et al. Biogeosciences, 2025 Evapotranspiration (ET) is an important variable for analyzing ecosystems, biophysical processes, and drought-related changes in the soil–plant–atmosphere system. In this study, we evaluated freely available ET products from satellite remote sensing (i.e., the Moderate resolution Imaging Spectroradiometer, MODIS; the ESA's Spinning Enhanced Visible and Infrared Imager, SEVIRI; and the Global Land Evaporation Amsterdam model, GLEAM) as well as modeling and reanalysis (i.e., the land component of the Earth system modeling product European Re-Analysis, ERA5-land, and Global Land Data Assimilation System version 2, GLDAS-2) together with in situ observations at eight Integrated Carbon Observation System (ICOS) stations across central Europe between 2017 and 2020. The land cover at the selected ICOS stations ranged from deciduous broad-leaf forests, evergreen needle-leaf forests, and mixed forests to agriculture. Trends in ET were analyzed together with soil moisture (SM) from the Soil Moisture Active Passive (SMAP) mission and the water vapor pressure deficit (VPD) from FLUXNET field measurements over 4 years, including a severe summer drought in 2018 and contrasting wet conditions in 2017. The analyses revealed the increased atmospheric aridity and decreased water supply for plant transpiration under drought conditions, showing that ET was generally lower and VPD higher in 2018 compared to in 2017. Across the study period, results indicate that during moisture-limited drought years, ET strongly decreases due to decreasing SM and increasing VPD. However, during normal or rather-wet years when SM is not limited, ET is mainly controlled by VPD and, hence, the atmospheric demand. The comparison of the different ET products based on time series, statistics, and extended triple collocation (ETC) shows generally good agreement, with ETC correlations between 0.39 and 0.99, as well as root-mean-square errors lower than 1.07 mm d−1. The greatest deviations were found at the agricultural managed sites Selhausen (Germany) and Bilos (France), with the former also showing the highest potential dependencies (error cross-correlation (ECC)) between the ET products (up to 7.6 and outside the acceptable range of −0.5 < ECC < 0.5). Thus, our results indicate that ET products differ most at stations with spatiotemporally varying land cover conditions (a variety of crops over growing periods and between seasons). This is because complex heterogeneity in land cover complicates the estimation of ET, while ET products agree well at evergreen needle-leaf stations with fewer temporal changes throughout the year and between years. The ET products from SEVIRI, ERA5-land, and GLEAM performed best when compared to ICOS observations, with either the lowest errors or the highest correlations.
Deep Learning Identification of the Governing Equation for Water Flow in Heterogeneous Soils From Data Wenxiang Song, Liangsheng Shi, Leilei He, Yuanyuan Zha, Xiaolong Hu, et al. Water Resources Research, 2025 Despite the remarkable advances in using deep learning for describing and predicting soil water flow, these models inherently cannot deepen our understanding of its underlying physical mechanisms as they are black‐box approaches. To address this issue, a novel data‐driven equation discovery approach has recently been widely used to facilitate scientific discovery in geoscience disciplines, including soil hydrology. However, due to the inherent complexity of soils, current data‐driven discovery approaches cannot deal with heterogeneous soil scenarios. In this study, we present a new group sparse regression theory and a deep learning framework to extend previous studies to be able to identify the governing equations for soil water flow in heterogeneous soils from observational data. Specifically, we focus on discovering equations from only time series of volumetric soil water content data, which are easily accessible. To accommodate it, the underlying assumption of the generalized soil‐water content‐based governing equation is utilized, and a coarse‐grained group sparsity theory is developed. Furthermore, we incorporate the proposed group sparse regression into a new deep‐learning framework: Extended‐DeepGS (Extended Deep‐learning‐based Group Sparsity). Through deep‐learning identification, it realizes simultaneous reconstructions of soil moisture dynamics and governing equations. A series of comprehensive numerical experiments are designed and conducted to test the performance of the theory and framework, and the results show its robustness. We also summarize the potential effects of soil heterogeneity on the discovery of equations. Finally, we discuss the limitations of the approach, which may inform future developments.
