*2.2. Real-World Model Fitting*

Water supply for the westside of the San Joaquin River Basin (SJRB) is provided by a water agency (e.g., United States Bureau of Reclamation) to the two westside subareas (Grasslands and the North-West Side subareas), according to water contracts negotiated between the water agency and individual water districts within each subarea. The individual water districts, in turn, allocate and distribute water supply according to agreements made with agricultural producers within each subarea. Water supply to subareas on the eastside of the SJRB derives largely from snowpack runoff from the Sierra Nevada Mountain Range, stored in downstream reservoirs along each major San Joaquin River (SJR) tributary. A state government water quality regulator (such as the California State Water Board), with the Regional Water Quality Control Board as enforcer, sets salt load objectives for the Basin in accordance with a Total Maximum Daily Load (TMDL) allocation developed by the Environmental Protection Agency for the basin. The load-based TMDL was further refined to develop subarea-level salt load allocations that take account of different water year hydrology. The conservative nature of the TMDL computation that utilizes the lowest 10% average low-flow condition resulted in allocations that were unattainable without major impact to the agricultural economy in each subarea. Hence, the initial TMDL allocations were replaced by concentration objectives, based on a 30-day running average electrical conductivity (EC), for the most downstream monitoring location on the SJR, Vernalis. A concentration objective allows agricultural producers and other salinity dischargers to utilize more of the available salt load assimilative capacity in the SJR. This initial compliance-monitoring objective has been supplemented with two additional upstream salinity objectives, ostensibly to protect the water quality of agricultural diversions made by westside agricultural producers. These additional salinity objectives are set at 1550 uS/cm year-round, as opposed to the 1000 uS/cm non-irrigation season, and 700 uS/cm objective set at Vernalis. The regulator suggested a number of approaches by which the original salinity load allocations, under the TMDL, might form the basis for salt load reduction strategies or cost allocation in situations where these various salt load objectives were violated.

The salinity load (mass of salt from all producers which is calculated by summing the product of drainage volume and salt concentration from each producer) produced on subarea *n* is the result of the return flows (drainage) from the agricultural activity of all producers, such that ∑*kn SnQn* ≤ *Sn*. There is no practical way that the regulator could equitably assign salt pollution levels to the individual agricultural producers or enforce this regulation at a reasonable cost to individual agricultural producers. Therefore, the regulator has chosen to allow stakeholders to internally govern the strategies to attain and abide by river EC objectives, while encouraging stakeholders to consider the subarea as the organizing entity for stakeholder action. Stakeholder compliance is monitored by the Regional Board using data supplied by state and federal water agencies.

To maintain compliance the agricultural producers can dynamically allocate salt loads to each subarea given that available salt load assimilative capacity at each compliance station is the product of the total assimilative capacity (defined by the current flow multiplied by the EC objective) and the current salt load in the river.

The monthly salt load cap can be calculated for each subarea individually based on the calculated TMDL allocations and the current salt loading to the river from each subarea (measured in terms of tons of salt: *SL* = *d* [salt concentration, *S*; volume, *Q*]), where *SL* is salt load (In the San Joaquin River the current TMDL criterion is a 30-day running average salt concentration that is multiplied by a monthly design flow to determine allowable salt loading). Using this stakeholder-maintained salinity load cap approach subareas would pay a fine (*F)* to the regulator, which could be a price per unit of salt load above the cap or some other equitable formula for dividing the fine amongst stakeholders. *F* can be specific to each subarea or similar for all subareas (see [23], for critique on uniform tax). *F* is then equitably distributed according to some formula (by land area, drainage volume, incremental salt load etc.) among all *Kn* agricultural producers in the different subareas (or by water user associations/districts in each subarea). We assume, for simplicity, that since we have a non-point source salinity management problem where the exact source of salt is not known, the most straight-forward and cost-effective initial approach to distribute *F* is to divide it equally per acre of land in production, or per acre–foot of irrigation water supply delivered. These initial approaches ignore the fact that some crops are associated with higher drainage return flow volumes and that subsurface drainage return flow salt loads may be poorly correlated with irrigation applications. Alternative allocation formulas may be relevant and will be considered in the empirical model. We assume that the (hypothetical) subarea manager (While there is no actual subarea manager, it is assumed that the model allocations are respected by the individual farmers and other decision makers at the water district level) has the authority or power to impose these allocations of river salt load assimilative capacity. We also do not want to set an optimal level for *F*, but rather take *F* as given in the empirical analysis. We will use several levels of *F* in a sensitivity analysis to evaluate the effect of *F* on the behavior of the agricultural decisionmakers at the subarea level.

An additional consideration is in the temporal administration of fines and fee schedules, which has a bearing on the design of a decision support system to aid the subarea manager to orchestrates stakeholder responses to potential violations of the river salinity objectives. An approach that attempts to respond to potential exceedance of salt load assimilative capacity at each compliance site in real-time would require model simulation tools that ran on a monthly timestep at a minimum. An optimization model would choose between available salt load reduction strategies, purchase of available water supply for dilution purposes, or payment of fines each month. Alternatively, accounting could be

postponed until the end of each year and fines imposed retroactively. The latter strategy would rely on uncertainty and the fear of a potential exceedance to motivate compliance. However, the decision tool needed to support this strategy could be simplified to operate on an annual timestep.
