Soil moisture product consistency for operational drought monitoring in Europe

Abstract

The roadmap to enable operational soil moisture (SM) monitoring for meteorologic and hydrological early warning depends on the capabilities of the available remote sensing and modelling products. Since each type of soil moisture product shows specific strengths and limitations due to their technical restrictions over certain environments, the detection of impactful anomalies across a wide range of conditions and scales is often challenging and incomplete without a combination of complemental data types of sufficient resolution, revisit time and coverage. This study evaluates the capabilities of SM products of different nature and their compatibility for combination, with special attention to their uncertainties in spatial consistency and in residual trends. While the first has been often revisited to validate remote sensing and modelling products against in situ data, the last is often overlooked in studies addressing SM changes despite its potential to disrupt the outcomes. To meet the demands of operational monitoring this study evaluated three SM products: (1) the Satellite Application Facility on Support to Operational Hydrology and Water Management (H SAF) active Advanced SCATterometer (ASCAT)-derived dataset, (2) the passive subset of the European Space Agency (ESA) – Climate Change Initiative (CCIp), and (3) the modelled dataset from the European Drought Observatory (EDO). The analysis was carried out over Europe in the period 2007–2022 at 10 d temporal scales and 5 km × 5 km spatial sampling. First, Pearson’s correlation coefficient (R) is used to measure the correspondence between H120, H121, CCIp and EDO SM products. Then triplets of the active, passive and model-based products are applied triple collocation analysis (TCA) to assess their performance based on TCA metrics such as the correlation, error variance, sensitivity and signal-to-noise ratio. We obtained that these popular well-validated datasets are increasingly capable in view of the notable TCA scores obtained but still subject to patches of spatial inconsistency and residual trends when compared against in situ SM data of the International Soil Moisture Network (ISMN). These uncertainties have minimal impact on drought monitoring in most of Europe, except in snow prone regions and for the assessment of long-term soil moisture trends used to design climate adaptation policies. Furthermore, each type of soil moisture product prevails in terms of triple collocation scores over the others under specific environmental conditions of the European continent. In view of the synergies shown by the active and passive remote sensing and the modelled SM estimates, two merged products are proposed and tested against the in situ data. The merging of the products is conducted by combining the various products based on weights calculated proportionally to the R_TCA scores of the triplets equalized in dynamic range matching their cumulative distribution functions. Results indicate that combining H SAF ASCAT, CCIp and EDO equals or surpasses the spatial and temporal consistency of the individual SM products alone, even when only the near-real-time products of H SAF ASCAT and EDO are combined. The evaluation of the trends of the individual products also indicates that small residual trends remain despite the improved filtering of the uncertainties, but given their differing sign of the trend, once combined into merged products can provide improved temporal stability of the series. Thus, merging remote sensing and modelled SM products enhances spatial consistency, resolution, temporal coverage and near-real-time capabilities for better European-scale drought monitoring, strengthening the early warning and risk management systems devoted to improving societal and environmental resilience.