Assessing field scale spatiotemporal heterogeneity in salinity dynamics using aerial data assimilation

Soil moisture and salinity shifts within the root zone can significantly alter crop yields. Thus, spatiotemporal dynamics of these parameters are essential for precision crop management. Airborne or spaceborne earth observation methods based on vegetation and soil observations have sometimes been used with limited success to indirectly understand parameters like soil salinity. These datasets lack spatiotemporal resolution to discern fieldscale heterogeneities, and estimates' accuracy is poor. A Metropolis-Hasting Markov Chain-Monte Carlo (MCMC) based method was developed to estimate field-scale soil salinity by assimilating estimated evapotranspiration (ET) data obtained from aerial canopy temperature sensing with ET outputs from a one-dimensional soil-water transport model. By aligning the two estimated ET values, we inferred anticipated soil salinity levels in a mature pecan orchard (28,951 m2). Our results aligned closely with in-situ measurements with a spatial cross-correlation more than 0.86 and highlighted the expected heterogeneities and nonlinearities. This research offers an approach to refine the current state-of-the-art crop models by accounting for field scale heterogeneities using remotely sensed data. This assimilation method will pave the way for a more inclusive agricultural system modeling that can infer critical but hard-to-measure soil properties from easier-to-obtain remotely sensed datasets. Though this paper concentrates on aerial observations, we anticipate similar methods can be used for satellite-based imagery, especially those with high spatial, temporal, and spectral resolutions.