Forecast of Natural Aquifer Discharge Using a Data‐Driven, Statistical Approach

Abstract

In the Western United States, demand for water is often out of balance with limited water supplies. This has led to extensive water rights conflict and litigation. A tool that can reliably forecast natural aquifer discharge months ahead of peak water demand could help water practitioners and managers by providing advanced knowledge of potential water-right mitigation requirements. The timing and magnitude of natural aquifer discharge from the Eastern Snake Plain Aquifer (ESPA) in southern Idaho is accurately forecast 4 months ahead of the peak water demand, which occurs annually in July. An ARIMA time-series model with exogenous predictors (ARIMAX model) was used to develop the forecast. The ARIMAX model fit to a set of training data was assessed using Akaike's information criterion to select the optimal model that forecasts aquifer discharge, given the previous year's discharge and values of the predictor variables. Model performance was assessed by application of the model to a validation subset of data. The Nash-Sutcliffe efficiency for model predictions made on the validation set was 0.57. The predictor variables used in our forecast represent the major recharge and discharge components of the ESPA water budget, including variables that reflect overall water supply and important aspects of water administration and management. Coefficients of variation on the regression coefficients for streamflow and irrigation diversions were all much less than 0.5, indicating that these variables are strong predictors. The model with the highest AIC weight included streamflow, two irrigation diversion variables, and storage.

Fractal nature of groundwater level fluctuations affected by riparian zone vegetation water use and river stage variations

“…5 employed detrended fluctuation analysis (DFA) to quantify fractal dynamics of groundwater systems. The complex scaling behavior still remains a challenge, especially when it changes with the temporal and/or spatial scales, in the prediction and quantification of subsurface processes 3,6 . Generally, the fractal structure of the groundwater level is determined by a power law exponent based on the assumption that the scaling is independent of space and time in DFA.…”

Temporal Scaling Analytical Method to Identify Multi‐Fractionality in Groundwater Head Fluctuations

“…The scaling characteristics embedded in the groundwater level fluctuations may reflect the critical hydrologic features of groundwater. Quantification of the groundwater process and forecast of groundwater‐flow regime, especially their temporal and spatial variations, remain a challenge in subsurface hydrology (Zhang and Schilling ; Boggs et al ). The natural processes and their characteristics can evolve with scales, which cannot be reliably characterized using a fixed mathematical format.…”

Forecasting the Probability of Future Groundwater Levels Declining Below Specified Low Thresholds in the Conterminous U.S.

“…Boggs et al . () used autoregressive time‐series modeling for forecasting aquifer discharge in southern Idaho. Changnon et al .…”