An analysis of the effects of retiring irrigation pumpage in the San Pedro riparian national conservation area, Cochise county, Arizona

Abstract

A seasonal groundwater model was developed to simulate fluxes and head distributions with periodic boundary conditions within the San Pedro Riparian National Conservation Area (SPRNCA) in southeastern Arizona. This model incorporated a seasonal approach for the period 1940-1995. Two years were used to simulate streamflow, 1990 and 1995. The model, as currently calibrated, does not accurately reproduce observed baseflow conditions in the San Pedro River and simulates an exaggerated effect of retiring irrigation within the SPRNCA. The model simulated increased baseflows while the observed baseflows declined at the USGS Charleston stream gage, though increases in baseflow contributions between Hereford Bridge and Lewis Springs have been reported. The original (Corell, et al., 1996) model and the seasonal transient model suffer from over- estimation of discharge from the floodplain aquifer to the San Pedro river, as well as errors in the seasonal transient model's simulation of riparian ET, and seasonal variations in stream conductance. These problems precluded the seasonal transient model from replicating the observed baseflows in the San Pedro river at the Charleston bridge, however, the results of the simulation are thought to be qualitatively indicative of changes in the flow system resulting from the retirement of irrigated agriculture in the San Pedro Riparian National Conservation Area. Possible sources for this problem include replacement of irrigation stresses by the expansion of cones of depression more distant from the river, overestimation of mountain front recharge, poor baseflow estimates and evapotransipration calculations from the stream gages at Charleston and Palominas, and the effects of a recently discovered silt -clay body that may dampen the speed of the rivers response to changes in stress. Additional efforts to re- calibrate the model, taking these areas into account, should provide better simulated baseflow values of the observed data.