Test Data for the Implementation of the MIEV Framework

This test dataset comprises four complementary datasets: combined_data, combined_mi_mtc, sm_data, and nearest_grid, which are required to implement and evaluate the Mutual Information Error Variance (MIEV) based surface soil moisture (SSM) fusion framework. The dataset combined_data contains gridded daily soil moisture observations over India for the period from 2016 to 2022, derived from three satellite products: SMAP, SMOS, and AMSR. The datasets are organized in a table that contains daily SSM observations, their corresponding spatial coordinates (lat, lon), and temporal information (time). The dataset combined_mi_mtc contains the spatially distributed Mutual Information (MI) and error variance values at each grid, which are required for the MIEV weighting scheme. The MI information characterizes the relationship between the individual soil-moisture products and the reference precipitation, while the error variances are obtained using the SPAR-TC (Spatially Representative Triple Collocation) approach. These parameters are used to derive the relative weights assigned to SMAP, SMOS, and AMSR during the fusion process. The dataset sm_data contains in situ, station-based soil moisture observations used as references for evaluating the fused soil moisture estimates. Each station is represented as a separate variable in the table. A user-defined subset of these stations is used as the optimization set to determine the optimal MIEV weighting exponent (n), while the remaining stations are used to generate and evaluate the final fused soil-moisture estimates. The dataset nearest_grid contains the spatial correspondence between each in situ station and its nearest gridded satellite soil moisture grid cell. It provides the station name (station) and the corresponding grid-cell latitude (grid_lat) and longitude (grid_lon). These datasets will be used to implement the complete MIEV workflow, where the optimal weighting exponent (n) is identified using selected optimization stations, and the optimized fusion scheme is subsequently applied to the remaining stations.