The vadose zone's physical properties significantly influence diverse hydrological and ecological activities across extensive spatiotemporal domains. Within agricultural contexts, fluctuations in soil moisture and salinity within this zone are pivotal in altering the yield of crops in a field. Remote sensing, utilizing satellite and aerial techniques, facilitates the non-intrusive investigation of various properties and processes by calculating indices (such as Normalized Difference Vegetation Index [NDVI] and Normalized Difference Red Edge [NDRE]) and Evapotranspiration (ET). While remote sensing proves effective for broad-scale dynamics, its spatial scales do not permit the identification and assessment of heterogeneity at the field scale, which is often dictated by hard-to-measure properties like soil salinity. This study illustrates a method to estimate soil salinity using data assimilation at a field scale. We integrated observed evapotranspiration (ET) data, captured from an Unmanned Aerial System (UAS)-based thermal sensor at various annual intervals, into a one-dimensional soil-water transport model. Aligning observed and predicted ET values facilitated the calculation of anticipated soil salinity. By creating individualized models for each tree (approximately 12m2), we computed expected variations in salinity across the entire field. The experiments were conducted in a mature Pecan orchard in El Paso, TX, serving as a case study. The results exhibited strong agreement with salinity levels identified by in-situ measurement methods, also revealing inherent heterogeneities and nonlinearities. The methodologies and outcomes from this research can enhance existing agricultural models by more accurately accounting for other challenging-to-estimate parameters like hydraulic conductivity, osmotic pressure, and ensuing salinity. This work may act as a foundational leap towards a systematic, physically based structure that assimilates non-optical properties using accessible remote sensing observations.
