Impacts of Irrigation on Soil Moisture Scaling Properties and Downscaling

Abstract:

Soil moisture (θ) exhibits high spatial variability due to the combined effect of natural and anthropogenic factors. Among the latter group, irrigation can introduce significant heterogeneity in the spatial variability of θ, thus modifying the statistical properties typically observed in natural landscapes. This, in turn, can affect the application of downscaling models of coarse satellite θ products based on the hypothesis of spatial homogeneity of θ distribution. In this study, the impact of irrigation on the scale invariance properties of θ and the application of a multifractal downscaling algorithm are analyzed using ground- and aircraft-based θ measurements from the National Airborne Field Experiments 2005 (NAFE05) and 2006 (NAFE06) campaigns conducted in two sites in Australia. After identifying irrigated areas through vegetation indices derived from Landsat 5 Thematic Mapper scenes, we investigate the presence of scale invariance from 32 km to 1 km in three scenarios, including (1) the original θ fields and in cases where θ in irrigated pixels was (2) replaced with missing data or (3) interpolated from neighboring pixels. We found that irrigation has a larger impact on the scale invariance properties in a large and compact agricultural district in the NAFE06 region, while it has a negligible influence on the sparser districts of NAFE05. The θ fields of scenario 3 are then used to calibrate a downscaling model based on spatially-homogeneous multifractal cascades as a function of coarse predictors. The model capability to reproduce the θ variability across scales is assessed by comparing ensembles of disaggregated field with the small-scale θ airborne observations and, for the first time, with ground θ measurements. Model performances are adequate in most cases in both experiments, although some deficiencies are found in regions with a larger presence of irrigated fields, suggesting the need to further refine the technique for detection of irrigated croplands. Results of this study provide insights on the spatial variability of θ in anthropogenic settings and are useful to refine satellite downscaling algorithms applied to Soil Moisture Ocean Salinity and Soil Moisture Active and Passive missions.