They used MODFLOW and particle tracking to trace the recharged water and concluded that the GIS analysis alone did not necessarily identify the most suitable sites. Likewise, Rahman et al. recommended that mathematical modeling should be combined with MCDA techniques to select optimal MAR locations even if a wide variety of thematic layers are considered in the MCDA. The criteria most frequently considered in site suitability analyses for MAR include intrinsic factors such as hydrogeology, topography, geomorphology, soil type, land use, and climate , as these represent main constraints on the groundwater recharge process . However, in view of the growing global population, urban sprawl, and water scarcity that affect both irrigation water supply and drinking water supply of groundwater-dependent rural communities, there is increasing recognition that integrated water resources management must consider socio-economic and socio-ecological criteria . The growing field of coupled natural human systems research in general, and socio-hydrology in particular , explicitly recognize the reciprocal interactions and co-evolution of coupled human-water systems. In California’s Central Valley , one of the most productive agricultural regions in the world, agricultural water use and climate-induced reduction of surface water supplies have led to severe groundwater overdraft in the past century. This has diminished access to clean,blackberries in containers reliable drinking water in rural communities .

This unsustainable and inequitable regional change has disproportionately impacted disadvantaged communities , who provide most of the farm labor to the agricultural sector , and perpetually decreased the economic viability and resilience of these communities to face hydro-climatic change . With the implementation of new groundwater legislation in California , MAR and especially Ag-MAR could play a central role in optimizing the use of surface water to stabilize depleted groundwater aquifers while addressing critical issues of drinking water supply and quality in rural communities. Targeted Ag-MAR in the well capture zones of rural communities could provide multiple hydrological, socio-economic, and socio-ecological benefits by increasing equitable access to groundwater resources for rural or impoverished communities while supporting the needs of a groundwater-dependent agricultural economy. The goal of this study is to delineate locations for targeted groundwater recharge on agricultural land with the potential to improve the groundwater supply in rural communities. Our study proposes a GIS-based MCDA methodology that combines biophysical and socio-economic data with groundwater modeling and particle tracking to identify and prioritize suitable Ag-MAR locations for multi-benefit recharge. In contrast to previous studies which tend to first focus on-site identification and then on groundwater benefits of recharge locations, our Ag-MAR site selection is spatially constrained to benefit domestic wells located in rural communities. We developed a GIS-based MCDA methodology with the specific objectives: to identify agricultural land parcels suitable for Ag-MAR, to map well capture zones in rural communities using a groundwater model and particle tracking, to estimate the vulnerability of rural communities to changes in groundwater supply, and to prioritize Ag-MAR sites of most benefit to rural communities based on community vulnerability.

The methodology framework was developed for California’s southern CV but could be useful to decision-makers and water resources managers in the efficient planning and management of Ag-MAR or MAR efforts worldwide.The southern CV of California , located between 35° and 37°N and 118° and 121°W, supports almost half of California’s $50 billion agricultural economy . The main agricultural commodities produced in the southern CV are table grapes, milk, almonds, alfalfa, pistachio and corn . Agriculture is the largest private employer in the region, with farm employment accounting for 25%–33% of all jobs . The area has a Mediterranean climate with most precipitation falling as rain or snow between November and April. Annual precipitation totals less than 200 mm in the valley and over 1,000 mm in the Sierra Nevada mountains and mean annual air temperature ranges between 17°C and 19.5°C across the valley floor. The Kings, Kern, Tule, and Kaweah rivers are major tributaries to the southern CV providing surface water to the region, most of which is stored in four major reservoirs for consumptive use during the dry summer season . High-magnitude flow and flood flow events, defined as flows above the 90th percentile of the full record, occur, on average, 2–5 years out of 10 years in the southern CV, creating a rare but important source of water for MAR . Kocis and Dahlke estimated that during years with high-magnitude flow, on average, 1.6 km3 of water is available in the San Joaquin Valley including the Tulare Lake Basin between November and April. Groundwater in the southern CV accounts for more than one-third of the water consumption, of which 68% is used for agriculture. Groundwater aquifers in the region have an average annual overdraft rate on the order of 2.2 km3 /year to 4.2 km3 . Higher rates were observed during the 2012–2016 drought and as a result of increasing groundwater use associated with the expansion of high value, perennial tree, and vine crops .

One-third of the local residents in the southern CV rely on domestic wells for drinking water, which tend to be shallower and withdraw less water than agricultural wells . Many domestic wells are compromised due to lowering water tables and contamination with nitrate, metals, and metalloids from agricultural activities and solvents and other chemicals from industrial activities . The four counties spanning the southern CV have some of the lowest median household incomes in the state, $47,518–$53,869 per year compared to the state’s median of $75,277 , and some of the highest poverty rates . Due to the lower income levels generally found in the southern CV, most communities meet the California Proposition 84 definition of a DAC—a community with a median household income of less than 80% of the statewide average household income . Approximately 353 of the 530 communities in the southern CV are considered DACs . In this study, we focus on 288 DACs, hereon called rural communities, located within the valley floor of the southern CV .Delineation of Ag-MAR areas that could improve drinking water supply in rural communities and estimation of community vulnerability to groundwater supply change is a multi-objective and multi-criteria problem that requires an understanding of the regional social-ecological-hydrological system . For this study, a total of 13 biophysical, hydrological, and socio-economic geospatial data were compiled . All data used in this study are from or representative of the 2012–2016 drought period. In addition to the biophysical data frequently used in MAR site selection , this study uses surface water conveyance infrastructure and groundwater flow fields to identify agricultural land parcels suitable for multi-benefit Ag-MAR and well source areas . To estimate the vulnerability of rural communities to change in groundwater supply,blackberry container pesticide applications, land subsidence, and U.S. Census data are combined. Figure 3 displays the GIS-based MCDA framework. All thematic layers were processed and integrated using the ArcGIS 10.6.1 The Soil Agricultural Groundwater Banking Index was used to identify land parcels suitable for Ag-MAR based on soil suitability . SAGBI considers five factors to accommodate groundwater recharge while maintaining soil health, crop growth, recharge efficiency, or groundwater quality . SAGBI ranks soils on a six-class scale ranging from Excellent to Very Poor. For this study, only agricultural areas with soil ranked as Excellent, Good, or Moderately Good were selected as suitable AgMAR locations based on recommendations provided by Dahlke et al. .Land in the southern CV is mainly used for agriculture , with only minimal urban settlements and riparian land cover remaining . The major crops grown in the southern CV are deciduous fruit and nut tree crops , field crops , and vineyards . Because prolonged flooding can cause waterlogging and anoxic conditions in the root zone, which can promote plant diseases and other pests , implementing Ag-MAR on agricultural land planted with perennial crops can increase the risk of yield loss. However, little research exists on crop tolerance to prolonged flooding conditions created by Ag-MAR. Preliminary research shows that vineyards and alfalfa show no significant yield penalties after controlled flooding from the end of winter to early spring when the fields overlay highly permeable soils . Thus, land parcels planted with vineyards, pasture , idle lands , grain and hay crops , and field crops, which likely are fallowed in the wet season, were deemed suitable for Ag-MAR.Geospatial data of existing surface water conveyance infrastructure were provided by nine water management agencies . For the remaining surface water districts, conveyance infrastructure was digitized from publicly available maps and aerial images. In areas where surface water conveyance infrastructure was visible in air photographs, features were digitized at a scale of 1:10,000. GIS data of larger conveyance infrastructure were obtained from the California Open Data Portal . All surface water conveyance data were classified into the following conveyance types based on district information: recharge basins, large canals, canals, drains, creeks, ditches, pipelines , and turnouts.

Surface water conveyance types were classified based on their accessibility for easier or less costly construction of surface water diversion points or extensions of existing infrastructure. The resulting classification used in the analysis was ditches, drains, creeks, pipelines, and canals, each comprising 22%, 1%, 10%, 25%, and 38% of the drainage network, respectively . Unique buffers were placed around each conveyance type based on typical distances measured between diversion points and conveyance sources . Distances were estimated from public surface water district maps and aerial images. Land parcels were prioritized based on which buffer they intersected.Groundwater well capture zones are defined as the areal extent and volumetric portion of a groundwater system that contribute discharge to a particular well . Captures zones were derived for all domestic wells in rural communities using the California Department of Water Resources’ Online System for Well Completion Reports and the California Central Valley Groundwater-Surface Water Simulation Model .C2VSim is an integrated surface water-groundwater model based on the finite element code of the Integrated Water Flow Model capable of accounting for reservoir deliveries, stream flow, stream diversions, canal distribution systems, irrigation, runoff, crop water uses, vadose zone processes, and groundwater-surface water-irrigated landscape interactions typical of irrigated agricultural basins . The model domain covers the entire CV alluvial aquifer. More information on the model development, calibration and validation, and application for Ag-MAR can be found in Brush et al. , Brush and Dogrul , and Kourakos et al. . The fine grid model version was used to extract a quasi steady-state representation of the regional groundwater flow field , taken as the mean of the monthly flow fields from October 2005 to September 2015, to best inform recharge efforts.In the State of California, construction of a new well must be reported to OSWCR. Well locations in OSWCR are reported at a 2.6 km2 resolution by stating the centroid coordinates of the 2.6 km2 Public Land Survey System section a well is located within . For this study, well logs of 27,482 domestic wells located within the study area were downloaded from OSWCR. Wells located within a 1.61 km radius of any of the 288 rural communities were extracted, reducing the number of domestic wells to 7,673 . Well logs were screened for completeness of the following well construction information: well depth, depth to the top of the well screen, and depth to the bottom of the well screen . For wells with incomplete records, well status modeling was used to impute missing information, specifically the depth to the top of the well screen and the submergence depth of the well pump .Investigation and mapping of factors that increase vulnerability to or probability of natural and socio-economic hazards can provide a useful way to prioritize where efforts should be focused . To determine community vulnerability to change in groundwater supply, data describing domestic well failures, domestic well reliance, pesticide applications, land subsidence, and socio-economic factors were used.A thematic layer describing “domestic well failures” was generated using domestic well construction logs, groundwater table information, and self-reported drinking water supply shortages . The California Household Water Supply Shortage Reporting System contains information on self-reported household drinking water supply shortages due to well failure, well under performance, or loss of surface water supply. During the 2012–2016 drought, 867 wells were reported dry within our 288 rural communities. However, this is assumed to be an underestimation of the actual water supply shortages experienced during the 2012–2016 drought, since reporting is optional and many rural communities likely lack access to these tools.Submergence of the well screen and pump intake are desired to ensure proper well function .