FORECASTING AQUIFER DISCHARGE USING A NUMERICAL MODEL AND CONVOLUTION

Aquifer discharge that occurs at some distance from an aquifer stress is not immediate. The discharge is lagged in time, creating an opportunity to develop a discharge forecast using knowledge gleaned from the nature of the stress. We developed a unique aquifer discharge forecasting procedure to meet two primary research objectives, 1) to quantify the proportions of aquifer discharge attributable to specific recharge and discharge components, and 2) to evaluate the efficacy of a one-year, deterministic forecast model based on initial aquifer heads and anticipated aquifer recharge and discharge. Some western states require groundwater users to compensate senior spring and surface water users for depleted flows (for example, Oregon Administrative Rules Chapter 690, 505 and 521; Idaho Administrative Procedure 37 3 11; Colorado Revised Statutes Sections 37-92-101 to 602). Reliable forecasts of aquifer discharge will benefit both water users and regulators by providing advanced identification of mitigation requirements and planning data for an appropriate course of future action.

We develop an aquifer discharge forecast and demonstrate important features of the procedure by presenting an application to the Eastern Snake Plain Aquifer (ESPA) in Idaho. We use an existing numerical model to obtain the aquifer discharge response from individual recharge components. The responses of these individual impacts, in the form of response ratios, are combined using convolution in a spreadsheet to generate an aquifer discharge forecast. The results indicate that an aquifer discharge forecast can be developed using only water supply projected effects on irrigation recharge at two large irrigation entities, aquifer pumping, and the effects of aquifer heads (initial conditions) at the time the forecast is generated. Analytical models suggest that ESPA annual discharge variability is caused by aquifer stresses no more than approximately 50 km from the discharge location. Our forecast is consistent with the analytical results, with annual aquifer discharge variability caused by stresses no more than 60 km from the discharge location. We also estimate how our forecast will perform in the future using a jackknife procedure, which indicates the future performance of the forecast is good (Nash-Sutcliffe efficiency of 0.83).