HERMES - Hybrid Estimation and Remote sensing Monitoring of Evaporation and Soil moisture

Context and objectives

Large-scale agricultural and hydrological management requires data of terrestrial evaporation (E) and root-zone soil moisture (SM) with high spatial resolution and accuracy. However, neither E nor SM can be directly measured from satellites, and in situ measurements are sparse and insufficient. Consequently, retrieval models and process-based algorithms have been developed during the past decades to indirectly estimate these variables from satellite data, sometimes using in situ measurements for calibration. The Global Land Evaporation Amsterdam Model (GLEAM) is one of these state-of-the-art methods that has been widely applied at a global scale and coarse resolution for over a decade, and its global E and SM estimates have been used to analyse trends in the water cycle, explore land–atmospheric feedbacks, and evaluate climate model accuracy.  

Until recently, the arena of water management and agricultural applications was closed for GLEAM due to its coarse (0.25 degree) resolution. Thanks to the STEREO III ET–Sense (SR/02/377) project, a 1-km resolution, 2-year pilot (2018–2019) dataset of E and SM over continental Europe was produced through the incorporation of Sentinel data into GLEAM. Moreover, during the STEREO III ALBERI (SR/00/373) project, deep learning algorithms were trained within GLEAM to estimate vegetation stress, resulting in a novel GLEAM-Hybrid prototype. This new perspective enabled higher accuracy in the E and SM estimates at global scales, but still at the original coarse resolution. A second-generation GLEAM-Hybrid could offer a unique opportunity to respond to the demands of the water management and agriculture sectors, if high resolution satellite observations were assimilated in a seamless and parsimonious manner. 

HERMES will yield a first-of-its-kind, high-resolution, high accuracy, E and SM dataset across Europe and Africa. It will bridge towards actual stakeholder requirements in the fields of water and agricultural management by exploring the influence of irrigation on E and SM by assimilating Sentinel 1 backscatter data into a second-generation GLEAM-Hybrid. The interpretable hybrid framework will enable an unparalleled exploration of the drivers of E and SM in different ecosystems, with specific emphasis on agricultural drought and heatwaves. The final outputs will be disseminated through an interactive web mapping tool based on the concept of 'datacubes', which will be regularly updated to provide timely information of interest to a wide range of end-users. The project feeds into the activities under the umbrella of the European Space Agency (ESA) Digital Twin Earth and the Global Climate Observing System (GCOS) and brings new conceptual understanding on crucial hydro-climatic variables while pushing scientific boundaries at the interface between Earth Observation and AI.