Soil moisture products underestimated plant-relevant dry-down during the recent drought in Germany

In the last decades, a variety of soil moisture products from remote sensing and process-based modeling have been created to study the terrestrial water cycle on a large scale. While satellite-based products are only representative of the water content at the topmost soil surface, model-based products aim at overcoming this limitation by estimating the water content in the deeper soil. In order to map the water content in the course of droughts and to analyze plant water absorption and transpiration, the water content in their rooting depths is of particular interest but out of scope for satellite-based products. Ground-based cosmic-ray neutron sensors are able to estimate soil water content at depths from 0 to 20 or 70 cm, depending on the soil water content. Their data offer a promising reference for the vertical extrapolation of satellite-based soil moisture products. In most soil moisture product assessment studies, assessment metrics are usually provided as single values over a certain period of time. However, this disregards the temporal dynamics of the metrics and the underlying processes. Here, we analyze the temporal dynamics of biases of cutting-edge soil moisture products from remote sensing and process-based modeling, in order to assess their potential to monitor plant-available soil water content. As a reference, we use soil moisture estimations from different sites of the Cosmic-Ray Soil Moisture Observation System (COSMOS) in Germany, covering a six-year time span (2015–2020) that includes the drought of 2018. We found that the biases have an annual frequency with a peak in summer for all selected products. Distinct peaks in 2018 and 2019 are outstanding and show the underestimation of the dry-down in subsurface soil layers caused by the drought. Additionally, there is a positive trend of the biases, even across different depths of multi-layer model-based products. The results suggest that the biases during the 2018 drought and subsequent years are due to soil drying at depths that are both below the coverage of the satellite sensors and not captured by the models. This demonstrates that the dry-down during droughts cannot be replicated by the chosen satellite- and model-based soil moisture products. For the accurate estimation of plant-relevant soil water content during droughts, a careful assessment of soil moisture products along with ground-based measurements is necessary. Our findings serve as a basis for improving current soil moisture products.