COMBINING FIELD CHARACTERIZATION, STATISTICAL ANALYSIS, AND GIS TO REGIONALIZE LOCALIZED RECHARGE ESTIMATES IN THE SUBHUMID PLAINS OF CENTRAL KANSAS

A practical methodology for recharge characterization was developed based on several years of field-oriented research at 10 sites in the subhumid plains of central Kansas. This methodology combines the soil-water budget and water-table fluctuation methods on a relatively high temporal (storm-by-storm year-round) resolution. The estimated 9-year (1985-1993) average annual recharge was less than 58 mm/yr, representing less than 10% of the average annual rainfall, with a recharge range from 15 mm/yr during the drought year of 1988 to 178 mm/yr during the 1993 flood year. Most of the recharge occurs during the spring months. To regionalize these site-specific estimates, a methodology based on multiple stepwise regression analysis combined with classification and GIS overlay analyses was developed and implemented. The multiple regression analysis showed that the most influential variables were, in order of decreasing importance, 1) total annual precipitation, 2) average maximum springtime soil-profile water storage, 3) average shallowest springtime depth to water table, and 4) average springtime precipitation rate. Therefore, four GIS (ARC/INFO) data "layers" or coverages were constructed for the study region based on those four variables, and each such coverage was classified into the same number of data classes to avoid biasing the results. The normalized regression coefficients were employed to weigh the class rankings of each recharge-affecting variable. This approach resulted in recharge zonations which agreed well with the site recharge estimates. During the "Great Flood of 1993," when rainfall totals exceeded normal levels by 200% in the northern portion of the study region, the developed regionalization methodology was tested against such extreme conditions, and proved to be both practical (based on readily available or easily measurable data) and robust (reliable even under extreme conditions.) We concluded that the combination of multiple regression and GIS overlay analyses is a powerful and practical approach to regionalizing small samples of recharge estimates.