Assessing the Impact of Spatiotemporal Representation on Agricultural Water Accounting Using Satellite Data

Satellite-based water accounting is increasingly used to estimate agricultural water demand in regions facing growing pressure from climate variability, land-use change, and limited availability of in situ observations. However, most operational applications rely on spatially aggregated satellite products that implicitly assume homogeneous conditions within basins or irrigation districts, thereby overlooking the spatiotemporal structure of hydrometeorological variability and associated measurement errors. The implications of this simplification for agricultural water accounting outcomes remain insufficiently quantified.

This study evaluates how agricultural water accounting results differ when spatiotemporal variability is explicitly represented, compared to conventional approaches that apply satellite products without detailed spatial and temporal reconstruction. A comparative framework is developed and applied to the Actopan River Basin in Veracruz, Mexico, which supplies Irrigation District 035 La Antigua, a region of high agricultural relevance dominated by sugarcane cultivation. Satellite-derived precipitation and reference evapotranspiration products for the period 2018–2024 are analyzed under two contrasting methodologies: (i) a baseline approach using non-interpolated satellite data, and (ii) a high-resolution approach incorporating spatiotemporal interpolation and error characterization.

Results show that neglecting spatial and temporal variability leads to systematic differences in estimated water balance components (P–ET), with implications for the magnitude, timing, and spatial distribution of agricultural water demand. Incorporating spatiotemporal structure enables identification of localized deviations that are masked under aggregated representations and provides a more realistic basis for accounting of crop water use. The analysis further demonstrates how systematic spatial and temporal discrepancies can be characterized and learned to improve consistency in water accounting calculations.

The proposed framework highlights the importance of scale-aware methodologies in satellite-based agricultural water accounting and is transferable to data-scarce basins where decision-making increasingly depends on remotely sensed information.