**1. Introduction**

Irrigated agriculture in semi-arid regions typically produces drainage return flows with high salinity content. Tanji and Kielen [1], review the conditions of low precipitation and high evaporation in semi-arid regions that lead to a high level of salinity in the drainage return flows. They provide examples from several locations such as the Nile Delta in Egypt, the Aral Sea Basin, and the San Joaquin Valley of California. An additional important region affected by salinity and the salinity pollution of water ways is the Murry Darling River Basin in Australia (Hart et al. [2]). The authors indicate that the clearance of deep-rooted native vegetation for the development of dryland agriculture and the development of irrigation systems in the basin have resulted in more water now entering the groundwater systems, resulting in mobilization of salt to the land surface and to rivers. In these examples, the return flows are discharged to water bodies (the Nile River, the Aral Sea, the San Joaquin River, and the Murry Darling River) that could benefit from regulation aimed to minimize negative externalities in the form of damage to crops and the environment. When these negative externalities exceed certain thresholds, the regulator can respond by assessing fines or other means of encouraging compliance with water quality objectives. We have seen different types of regulators, such as a river basin authority, in charge of managing the quantity and quality of water in the basin. A river basin authority could be a state or federal entity with the authority to impose fines or restrict water allocations in the case of nonpoint

**Citation:** Dinar, A.; Quinn, N.W.T. Developing a Decision Support System for Regional Agricultural Nonpoint Salinity Pollution Management: Application to the San Joaquin River, California. *Water* **2022**, *14*, 2384. https://doi.org/10.3390/ w14152384

Academic Editor: Athanasios Loukas

Received: 24 April 2022 Accepted: 28 July 2022 Published: 1 August 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

source pollution. Baccour et al. [3] model the case of the Ebro River Basin in Spain, where regulations to control nonpoint source pollution of nitrates from livestock production are addressed, among other policy interventions, by restriction of water supply, imposed by the Ebro Basin Authority. Quinn [4], compares the performance of real-time, basin-scale salinity regulation in the San Joaquin River in California with those of the Hunter River Basin authority, Australia.

Some numerical simulation models can be configured to act as decision support tools, which provide alerts of potential violations of water quality objectives, and can assist in the development of schemas for creating incentives or assessing fines to encourage compliance [5,6]. Obropta et al. [5] developed a model to address hot spots for a water quality trading program intended to implement the total maximum daily load (TMDL) for phosphorus in the Non-Tidal Passaic River Basin in New Jersey. Zhang et al. [6] develop a model-based decision support system for supporting water quality management under multiple uncertainties. Such tools can have the added benefit of allowing equitable imposition of proposed incentives or fines on those polluters who bear the primary responsibility for the load exceedances. Similar decision support tools have been developed in other sectors and contexts. Ioannou et al. [7] designed a DSS to help managers in the process of decision making, in handling areas that have been burnt by forest fires, by running hypothetical (what-if) scenarios in order to achieve the best form of intervention in fireaffected regions of Greece. Makropoulos et al. [8] demonstrate the development and use of a DSS to facilitate the selection of bundles of water-saving strategies and technologies to support the delivery of integrated, sustainable water management in the UK. Rose et al. [9] identify factors affecting the selection and use of decision support tools by farmers and farm advisers in the UK for agricultural planning purposes.

Our paper presents an approach using a regional framework that enhances the utility of existing modeling tools (Watershed Analysis Risk Management Framework—WARMF), which are currently in use by practitioners in the San Joaquin Valley (Systech Water Resources Inc. [10]), to make forecasts of the salt load assimilative capacity of a major river. The river receives salt loads from catchments in the form of irrigation return flows that often exceed the river's salt load assimilative capacity. Uses of the modeling framework WARMF by practitioners are described by Quinn et al. [11], Fu et al. [12], and Quinn et al. [13], where WARMF features and performances are compared with other tools currently in use by practitioners.

The nonpoint source nature of agricultural salinity pollution poses a dual challenge for regulators by making it difficult to identify primary polluters, and to quantify pollution loads on a continuous basis. Not all drainage outlets can be monitored; therefore, calibrated simulation models play an important role in predicting pollutant loads under various permutations of hydrological and water quality inputs. Models allow alternative regulatory approaches, including schemes such as voluntary agreements and cap-and-trade in pollution permits to be evaluated, provided they can be adequately calibrated. Examples for such schemes are discussed and explained below [14–23].

Published literature on economic and regulatory aspects of nonpoint source pollution in irrigated agriculture highlights a variety of socio-political issues. These include the role of asymmetric information, value of information, effectiveness of policy interventions, and adoption of pollution reduction production practices. In an early work Griffin and Bromley [14], established a conceptual model for analyzing agricultural nonpoint pollution. An important aspect of pollution quantification is the representation of the biophysical processes linking production decisions to emission loads. Production decisions are reflected in the type and quantity of inputs in management practices and in local biophysical conditions.

An extension of the analysis in [14], was proposed in Shortle and Dunn [15], which included stochastic components in the pollution functions that arose from random natural processes as a means of addressing the lack of information about key biophysical processes. A review of various nonpoint source pollution control regulations (either incentives, taxes, or quotas) on inputs was also provided in Shortle and Dunn [15]. These are second-best interventions in the absence of direct measurements of polluter discharges. Shortle and Dunn [15], identified a reduction in the cost-effectiveness of these pollution control measures when applied uniformly across diverse agriculturally dominated subareas, which are heterogeneous in terms of water management practices and landscape characteristics, that can lead to different receiving water impact functions. To address such shortcomings, a costeffective approach was developed by Shortle at al. [16] and was used to determine the best single-input tax policy for nonpoint source pollution in agriculture. The authors examined the question of reducing nitrate leaching from lettuce fields in California. Larson et al. [17] argue that under the certain circumstances applied, irrigation water can be the easiest single input to regulate since nitrate loading to groundwater is directly related to soil leaching rates. However, for other contaminants such as salinity, salt loads in subsurface drainage return flows may not be well correlated with surface water applications since most of the salt captured by the sub-surface drains may originate from deeper layers in the aquifer rather than from infiltrating water. Considerations of transaction costs and other political, legal, or informational constraints for dealing with nonpoint source pollution regulation were presented in Ribaudo et al. [18]. Such considerations could be applied to achieve specific environmental goals in a cost-effective manner. The authors discussed the economic characteristics of five instruments that could be used to reduce agricultural nonpoint source pollution (economic incentives, standards, education, liability, and research).

Several authors [9–23], considered regulation that had a spatial component in the presence of heterogeneity instead of regionally uniform instruments. In these works, authors demonstrated that spatially uniform policies resulted in economic efficiency losses and reduction in welfare. Kolstad [19], modeled a two-pollutant economy and showed that when marginal costs and benefits become steeper, the inefficiency associated with undifferentiated regulation increases. Wu and Babcock [20], demonstrate, among other things, that a uniform tax on polluter farmers may result in some farmers not using the chemical, and a uniform standard may have no effect on low-input land. Doole [21], finds that because of the disparity in the slopes of abatement–cost curves across dairy farms in New Zealand, a differentiated policy is more cost-effective at the levels of regulation required to achieve key societal goals for improved water quality. Doole and Pannell [22], find that due to variation in nitrate baseline emissions and the slopes of abatement–cost curves among polluting dairy farms renders a differentiated policy which is less costly than a uniform standard in the Waikato Region of New Zealand. Finally, the work by Esteban and Albiac [23], demonstrated and quantified the welfare loss from a spatially uniform regulatory policy to reduce salinity pollution and the efficiency gains from different policy measures based on the same spatial characteristics, applied in the Ebro River Basin of Spain.

Very few studies consider joint management of the nonpoint source pollution in a regional setting, using cooperative arrangements and trade, including trade-in water rights/quotas and trade-in pollution disposal permits in a regional setting. Several examples from actual cases exist. The Murray Darling Basin Authority [24], initiated a basin-wide agreement, a joint work program designed for setting salt disposal permits based on historical loads, including a revised cost-sharing formula and salinity credit allocation shares for Victoria, New South Wales, South Australia, and the Commonwealth [25]. In the Hunter Basin of New South Wales, Australia [25,26], a scheme of salt permit discharges has been put into work. The main idea of this scheme was to permit discharge of salt loads only when there was available salt load assimilative capacity in the Hunter River that drains the Hunter Basin. Quinn [4], reviews how salt load discharges to the river were scheduled by quantity, time, and location based on stakeholder need and calculations of salt load assimilative capacity using a simple spreadsheet mass–balance model.

Increasing use of high-salinity water as an irrigation source could be a serious problem. Yaron et al. [27] analyzed the economic potential to address such a problem by cooperative settlements in Israel and calculated income distribution schemes for three farms, using cooperative game theory (GT) algorithms. Work by Dinar et al. [28] also applied cooperative GT to the regional use of irrigation water under scarcity and salinity. Their model addressed inter-farm cooperation in water use for irrigation and determined the optimal water quantity and quality mix for each water user in the region.

Several additional works that represent various efforts and methods include Nicholson et al. [29] who conducted a comprehensive assessment of decision support tools used by farmers, advisors, water managers, and policy makers across the European Union as an aid to meeting the EU Common Agricultural Policy objectives and targets. Development and use of a GIS-based decision support framework was suggested by Chowdary et al. [30], integrating field scale models of nonpoint source pollution processes for assessment of nonpoint source pollution measures of groundwater-irrigated areas in India. A GIS was used to represent the spatial variation in input data over the project area and to produce a map that displayed the output from the recharge and nitrogen balance models. Different strategies for water and fertilizer were evaluated using this framework to foster long-term sustainability of productive agriculture in large irrigation projects.

The work by Quinn [4], which uses monitoring, modeling, and information dissemination for salt management in the Hunter River Basin in Australia, was compared to a more model-intensive approach deployed in the San Joaquin River Basin in California. Decision support systems for these river basins were developed to achieve environmental compliance and to sustain irrigated agriculture in an equitable and socially and politically acceptable manner. In both basins, web-based stakeholder information dissemination was a key for the achievement of a high level of stakeholder involvement and the formulation of effective decision support salinity management tools. The paper also compared the opportunities and constraints of governing salinity management in the two basins as well as the integrated use of monitoring, modeling, and information technology to achieve project objectives.

In the present paper, we provide a scalable water quality simulation model and decision support tool for a regional water quality (salinity) management problem that incorporates water/irrigation regions, each serving several individual farmers. The model operates at the subarea level where each subarea has distinct features that include political and hydrologic boundaries and which recognize different accesses to water supply and drainage resources. These subareas have been recognized by the State of California water regulatory agency with jurisdiction over the project area. We highlight the role of top-down regulations as well as market-based arrangements that might form a basis for cap and trade in pollution permits. We compare and discuss the physical as well as the welfare consequences of various policy interventions. The combination of monitoring system networks and decision support frameworks are scalable—hence, the final work product can be applied at the individual stakeholder level or aggregated at the water district level. Institutional and managerial components of the schema would need to be separately developed. Regional cooperation in the form of a market for tradeable salinity pollution permits [16], would be a significant outcome of a future study and one that is facilitated by the unique application of the simulation modeling tool.

The paper develops as follows: First, we present the analytical model aimed to evaluate the various options for pollution control at the subarea (regional) levels, responses of individual dischargers to introduced regulations, and the allocation of joint costs and benefits among the salt-discharging regions. Next, we introduce a proposed empirical framework to be applied to the San Joaquin River in California, given existing model resources in use by regulatory agencies. Then, we define a subset of seven subareas within the region as the basis for the empirical application aimed to test the analytical model. Finally, we evaluate the results to expand the method to incorporate future cooperative strategies.
