Spatiotemporal evaluation of downscaled and native high-resolution satellite soil moisture products

Abstract

ESA’s Sentinel-1 mission has enabled global, operational monitoring of soil moisture (SM) at (sub)kilometer scales for over a decade. Simultaneously, spatial enhancement techniques for coarse-resolution SM products have emerged, producing SM estimates at comparable spatial scales. However, differences between these two approaches – particularly in their ability to represent true spatial features – remain poorly understood owing to the limitations of current evaluation strategies. Application-driven assessments provide indirect performance characterization relying on non-SM information, while in situ measurements are considered too sparse or granular for spatial validation. The recently proposed Point-Scale Downsampling (PSD) framework addresses this challenge by enabling relative spatiotemporal comparisons between fine-scale observations and coarse-scale benchmark data. However, its potential for assessing differences among fine-scale SM products has not yet been tested in a coordinated, round-robin evaluation. To address this gap, we apply PSD to six high-resolution satellite products: two directly retrieved from Sentinel-1 SAR observations and four derived from downscaling or data fusion methods based on coarse-resolution retrievals. We additionally assess temporal agreement with in situ networks and establish a temporal–spatial context using correlogram-based spatial variability metrics to test the assumption that greater fine-scale variability reflects increased SM detail. Our results indicate that while enhanced products exhibit lower spatial variability than SAR retrievals, they generally show stronger spatiotemporal agreement with in situ measurements. In contrast, SAR-based products tend to achieve lower performance scores across most networks, suggesting persistent challenges in accurate fine-scale parameterization. PSD provides a viable framework for comparative assessment, though its capacity for detailed stratification and explanation is limited.