A statistical modelling approach to site suitability of small agricultural reservoirs

Agricultural production increasingly relies on irrigation to withstand droughts, precipitation variability or support agricultural intensification. Small Agricultural Reservoirs (SmARs) can contribute to sustainable agricultural water management by providing additional water without increasing pressure on surface and groundwater resources. The construction of SmARs is usually subject to a phase of suitability analysis, which helps discern suitable places within a large area, before exploring the potential locations with major details. This task is traditionally performed using deductive approaches relying on multi-criteria decision analysis (MCDA), which are based on relevant macro criteria for the location of SmARs, often supported by hydrological modelling. In this work, we present a data-driven statistical modelling approach based on a large database of existing SmAR locations. Our empirical approach shows a better match with a validation sample, with about 71% of SmARs in very high suitability areas compared to about 5% for the MCDA for the Italian case study in Tuscany. Moreover, our approach provides suitability models for different SmAR sizes, aiding more detailed analysis.Our results can directly support high level suitability in Tuscany, while the proposed approach can be further extended and applied in different contexts, scales, and applications.