Integrating hydrological and economic modeling to assess the impacts of adaptation policies

Abstract

Numerical models play a vital role in representing geohydrological processes and informing the management of surface and subsurface water flows. Yet, these models have limitations, such as not being able to determine the behavior and responses of water users, and the resulting pressures on water bodies via technological, land and water allocation choices. Microeconomic models can aptly complement geohydrological models due to their ability to quantify land and water use choices. Several studies have seeked to combine the strengths of water and human system models using modular or holistic couplings. In a recent review, assess 198 integrated human and water systems models and find that in 88 of these models the integration focuses either on the surface water or the groundwater system. As shown by (Salmani et al.,2023), Simulating surface water or groundwater alone may not accurately represent water system dynamics, leading to important modeling errors that may cascade to human systems and lead to bias forecasts. This study develops a modular hydro-economic model that explicitly models surface water and groundwater systems. The water system is populated by SWAT+gwflow, which integrates the Soil Water Assessment Tool (SWAT+) with the groundwater module gwflow; while the human system is populated with a microeconomic positive mathematical programming (PMP) model that represents the behavior of irrigators. The proposed model is illustrated with an application to the overlapping Cega-Eresma-Adaja sub-basin and Arenales Aquifer in Spain. The model setup is implemented in two steps. At first, the PMP is calibrated for each Agricultural Water Demand Unit, the basic irrigation water use unit in Spain, using observed land and water use data and socioeconomic data for the period (2015). Then, the SWAT+gwflow model is calibrated for 43 subbasins and 1247 HRUs from 1990 to 2020. This model was calibrated and validated in 14 observation gauges and 29 observation wells to evaluate the streamflow and head of the aquifer. The model showed a Nash-Sutcliffe efficiency of 0.65-0.85 and coefficient of determination of 0.7-0.9 for all stations in the baseline, indicating good simulation. The simulated groundwater head showed good agreement with observed well data, with a mean absolute error of less than 0.5 m in the baseline and other scenarios. Moreover, the rivers were found to be heavily dependent on groundwater discharge to streams. Once the two models are calibrated, according to the river ministry policies derived from the historical droughts graph, a series of simulations in which irrigated crops land and water use are constrained across AWDUs are run in the PMP model in urgent or alert situations. The resultant crop portfolio in the PMP simulation is replicated in the SWAT+gwflow, and the water use and management practices updated to match those of the PMP. Finally, simulations are run with the SWAT+gwflow to assess the impact of land and water reallocations by irrigators on the surface water and groundwater systems. Results show that decreasing the amount of land and water used for irrigated crops can increase stream flow and lead to more normal conditions while increasing the portfolio of rainfed crops.