Inversion of irrigation from satellite soil moisture data with a model based on PrISM (Precipitation Inferred from Soil Moisture)

Accurate irrigation water management is crucial for maximizing crop yield and minimizing water waste. Remote sensing technology offers a promising solution for efficiently estimating irrigation water use at the field scale.In this study, we adapted the PrISM (Precipitation inferred from Soil Moisture) methodology to detect and estimate irrigation events from soil moisture remotely sensed data. PrISM is a well-known approach to correct precipitation estimates using soil moisture data. Its main application is to provide a near real-time corrected precipitation product. PrISM employs an antecedent precipitation index (API) formula coupled with a particle filter assimilation scheme for soil moisture.In this study, we adapted the PrISM methodology to estimate irrigation amounts from soil moisture. The methodology uses initial precipitation estimates and soil moisture profile to detect whenever water excess is present in the soil (not caused by precipitation) and estimates its amount, together with its uncertainty. The methodology does not need extensive calibration and it is adaptable to different spatial and temporal scales. A synthetic study was performed to investigate the effect of a degraded soil moisture signal in terms of temporal resolution (lowering the temporal sampling of the soil moisture time-series), spatial resolution (lowering the percentage of irrigated area in a pixel), and random noise (increasing RMSE values). Results from this study suggested that high spatial resolution is critical in order to avoid underestimation of irrigation amounts. Ideally, a field-level soil moisture (with more than 75% of the pixel irrigated) and a product with low RMSE (0.02 m3/m3) is required for precise estimations (in order to keep the error of annual cumulative irrigation below 20%). Temporal resolution has a lower impact, especially when an assumption on the frequency of irrigation events (deduced from the system of irrigation used at the field-level) is included in the algorithm.Consequently, the developed algorithm was applied to actual satellite soil moisture products at different spatial scales over the same area. Validation was performed using in situ data at the district level of Algerri-Balaguer from the study area in Catalunya, Spain, where ground-based irrigation amounts were available for various years. Additional validation was performed at the field-level at the Segarra-Garrigues irrigation district using in-situ data from a few fields where soil moisture profiles and irrigation amounts were continuously monitored. Our results suggest that PrISM can be used effectively to estimate irrigation from soil moisture remote sensing data and that this methodology could be potentially applied on a large scale, with the only limitation being the quality and spatial resolution of the satellite soil moisture product.