Downscaling the ESA CCI Soil Moisture: a new European dataset at 1 km for the period 2008-2020

Abstract

The European Space Agency (ESA) Climate Change Initiative (CCI) provides long-term surface soil moisture (SM) records with daily temporal resolution. However, the coarse spatial resolution of approximately 25 km limits their use in many hydrological applications, such as agricultural water management, drought monitoring, and rainfall-runoff response.

To address this constraint, we downscaled the CCI SM product to 0.01° (~ 1 km) using machine learning and a set of static and dynamic variables affecting the spatial organization of SM. In particular, datasets describing the vegetation status throughout time, as well as land cover class and soil and topographic attributes were fed into a Random Forest model.

Here, we will first present in detail the methodological framework that allowed us to generate the high-resolution dataset. Then, we will thouroughly evaluate its accuracy against in-situ measurements from across Europe, and further compare it to other SM products (e.g., from Sentinel-1). Finally, we will highlight the strengths and limitations of the downscaled SM dataset and discuss possible improvements.