The initial installation of the pump intake is usually above zt to minimize costs of screen installation and to maximize the capacity for useable water production . It was assumed that the cost of rehabilitating wells to alleviate well production losses caused by falling groundwater levels would be prohibitive for rural communities. Thus, the analysis assumed that pump intakes would remain at some depth above zt and wells would become inactive if the groundwater level dropped below the pump intake . Since the OSWCR contains no information on the pump intake depth, a submergence value, hs, was calibrated using the reported well failures as a validation data set. Submergence in this study is defined as the depth of the top of the well screen below the groundwater table. Depth to the groundwater table and estimated zt values were used to quantify changes in well status between Spring of 2011 and Fall of 2015 . Groundwater depths at each well location were extracted from interpolated seasonal groundwater levels spanning the entire shallow to semi-confined CV aquifer system . To calibrate the required submergence value, hs, zt values were compared to predrought and post drought groundwater levels to identify wells that became inactive as a result of groundwater level declines. Using the reported well failures in rural communities as validation data, the required pump submergence value was calibrated to be hs = 10 m for which the model estimated 923 well failures during the drought period, most of which were concentrated in the northeastern region .Many rural communities in the southern CV are not connected to municipal water supply systems and generally rely on a single water source, typically a groundwater well,planting blueberries in a pot which puts them at risk of water supply failures . Water supply connection density is a metric that describes the pressure exerted on community drinking water supply sources, given as the ratio of active public water supply sources to water supply connections in each community .

Lower values indicate a higher per capita reliance on active public water supply sources, indicating the community has lower water supply security. Communities that rely on a single public water supply source are especially vulnerable to shortages and contamination, as the failure of a single source compromises the community’s entire water supply. In the study area, 91 rural communities only have a single public water supply source of which more than 75% rely on groundwater. Communities solely reliant on unregulated domestic wells do not have any access to public water supply sources and as such, are the most vulnerable to shortages and contamination. The communities reliant on single or unregulated sources are concentrated in the northeastern and eastern regions of the southern CV.The California Department of Pesticide Regulation recognizes the following seven active ingredients contained in pesticides as a public health risk having the potential to pollute groundwater: atrazine, simazine, bromacil, diuron , prometon, bentazon, and norflurazon . Records of total annual application amounts of these active ingredients were obtained from the California Pesticide Information Portal for the year 2015. Values range between 0 and 1,024 kg within the study area , with higher loads concentrated in northeastern and eastern regions .Prolonged and unsustainable groundwater pumping causes severe settling or sinking of the land surface due to subsurface compaction of earth materials, known as land subsidence . Land subsidence rates estimated with InSAR technology between May 2015 and September 2016 was used in this analysis. The data reveal two major subsidence bowls in the northwestern and eastern regions of the southern CV and the development of a new hot spot between them . Land subsidence is of particular concern because it directly affects major surface water conveyance systems and threatens the integrity of shallow, domestic wells.Socio-economic parameters of poverty status, linguistic isolation, and educational attainment were selected as unique and complementary factors contributing to community vulnerability to change in groundwater supply . Socio-economic data were obtained by block group from the U.S. Census Bureau’s American Community Survey’s 5-year estimates for 2011–2015 and processed using the R library tidycensus . For each of the three parameters described below, demographic percentages were calculated for all block groups in the region. If multiple block groups intersected a community, an area-weighted average was calculated and the value was applied to each respective community. Poverty status is defined as the percentage of the population for whom the ratio of income to national poverty level in the previous 12 months was below one .

Poverty status is believed to contribute to community vulnerability as poorer households have less financial capacity to preemptively address or remediate water supply shortages .Households that have limited English-speaking capacity are to a lesser extent able to engage with administrative authorities to voice concerns or resolve problems, and thus have increased community vulnerability . Educational attainment is defined as the percentage of population over 25 years of age, who have completed some education above the high school level . Educational attainment can influence risk perception, skills and knowledge, and access to information and resources, hence less educated populations may be less empowered to prepare and recover from resource shortages .A GIS-based MCDA was used to combine the biophysical, hydrological, and social-ecological data listed in Table 1 to delineate and prioritize locations for multi-benefit Ag-MAR. An equal weighting scheme for thematic layers and proposed rankings of categorical features was adopted in this study following recommendations of Visser and based on the variability present in existing recharge mapping studies .Thematic layers “soil suitability for groundwater recharge,” “land use and land cover,” and “surface water conveyance infrastructure” were combined to assess the suitability of land parcels for Ag-MAR . Boolean criteria were used to restrict focus to soil types that allow percolation of surface water into groundwater aquifers, land use, and land cover types that show tolerance to prolonged flooding conditions, and land parcels that are near existing surface water conveyance infrastructure. The capture and source area of a groundwater well is dependent on the depth of the well, length of the screened section, and the groundwater flow field. A particle tracking algorithm using the Runge-Kutta-Fehlberg numerical method was implemented to identify the capture areas for all domestic wells within rural communities. The Runge-Kutta-Fehlberg uses a self-adaptive step procedure, where the step size is reduced as the curvature of the particle trajectory is increased . Using the quasi steady-state groundwater flow field extracted from CDWR’s C2VSim model, the particle tracking algorithm calculates the velocity and trajectory of a particle by interpolating the velocities between the nearest points of simulated groundwater heads in the model, then transporting the particles backward in time to determine their exit points using discrete steps of a predefined time length.

Well construction information from the OSWCR database and well status modeling , including the well location, depth of well screen, and screen length, are used for each domestic well within a rural community. More information on the parameters used in the particle tracking can be found in the Supplemental Materials.Although a wide variety of decision support tools are available for general surface and groundwater management and drinking water quality in California ; none of these tools provide information on mitigation or remediation options for chronic groundwater overdraft or contamination. This study is the first effort to systematically explore the potential for targeted Ag-MAR to directly improve the drinking water supply from groundwater in rural communities. In past decades, MAR has been used to achieve varying objectives , however, implementation of MAR is often limited by challenges of recharge water availability , locating suitable groundwater recharge zones, regulatory constraints, and funding obstacles . Ag-MAR overcomes many of these challenges due to low capital cost and permitting requirements ,raspberries in pots and with appropriate planning can be used to provide multiple benefits to a region including stabilized domestic and agricultural water supply, flood control, and climate change mitigation . However, Ag-MAR implementation in the southern CV might be constrained by the existing surface water conveyance capacity, which Hanak et al. deemed inadequate for capturing and moving high flows to suitable recharge locations. Conveyance capacity data were not available for this analysis, but according to Hanak et al. represents one of the major limitations for MAR implementation. In this study, almost 3,000 land parcels suitable for Ag-MAR ranging in size from 0.2 to 260 ha have been located within the well capture zones of rural communities. Of the 288 rural communities included in this analysis, 253 communities rely on groundwater as their main source of drinking water. However, suitable Ag-MAR parcels could only be identified within the capture zones of 149 of the 288 communities, 144 of which are reliant on groundwater for their drinking water supply. Most of the communities for which no nearby AgMAR parcels could be identified are located near large urban areas or near the CV rim, where topography and a lack of conveyance infrastructure prohibit Ag-MAR. A complex political and socio-economic environment around water governance in the region has historically prevented more inclusive water management but for these communities, other types of MAR , well head treatment, or incorporation into nearby public water supply systems might be the only options to improve the quantity and quality of drinking water supplies. For reference, 118 of the 288 communities studied have no access to public water supply sources but 56% of these communities are within the boundaries of existing public water supply systems.MAR site selection studies using GIS-based MCDA approaches have been developed in many regions across the world . The majority of these studies use slope, land use, geology and soil type as the main criteria for identifying MAR sites . Similarly, our study uses soil characteristics and land use as the main criteria to determine Ag-MAR site suitability, but differs from earlier studies in that we refine suitable sites by linking the GIS analysis with deterministic groundwater modeling and particle tracking to only select sites with potential to benefit the drinking water supply in rural communities.

The integration of groundwater modeling and particle tracking also ensured the inclusion of climate and hydrogeological data in the analysis. However, the groundwater modeling also introduced uncertainty in the estimated well capture zones, due to the spatio-temporal resolution of the model and because a quasi steady-state groundwater flow field was used for the particle tracking. The generalized groundwater flow field likely does not capture local spatio-temporal dynamics in the flow field caused by seasonal pumping, which can change or reverse some of the flow directions depicted in Figure 5. These seasonal dynamics should be considered in the final selection of Ag-MAR locations using field-level studies. In addition, in groundwater-dependent regions where an integrated surface water-groundwater model is not available, well capture zones may need to be derived from field observations. The Ag-MAR locations identified in this study relied on the integration of regionally specific data for the southern CV, but the methodology can be applied to other groundwater-dependent regions. To implement the Ag-MAR site suitability analysis, regional soil or geomorphology data can be used instead of SAGBI, and land use and surface water hydrology can be inferred from air photographs and satellite images. Similarly, data descriptive of the socio-economic status of rural communities in groundwater-dependent regions or adverse environmental effects of human activities and groundwater overdraft on rural populations can be substituted with locally available demographic data or remote sensing data , respectively. In regions where little geologic or physiographic data exists, growing availability of high-resolution remote sensing data of land surface and subsurface characteristics may be useful . Many previous MAR site suitability studies were conducted to inform sustainable groundwater management , to serve as guidelines and decision support for farmers and policy makers , or to raise general interest for MAR development . However, as showcased in this study, GIS-based MCDA can also be used to identify priority areas for intervention or disaster management if site suitability analysis is combined with vulnerability analysis . This combination can be particularly useful in water resources management because the outputs can provide easily interpretable visual information, help refine the spatial focus of the problem, support priority development, and allow for assessment of different management scenarios before field-level investigations begin.