This study was made to determine the accuracy of using readily available data with certain statistical techniques to estimate ground-water withdrawals in western Kansas. The data used in the investigation were from a sample of wells chosen from the total inventoried irrigation wells in nine counties in northwestern Kansas; they can be considered as being typical of the data generally available in western Kansas.
The hypothesis that each of three physical characteristics of the wells had the same distribution in the sample as in the total population was accepted at the 95 percent significance level. These characteristics (saturated thickness of the aquifer, depth to water below land surface, and reported well yield) were assumed to be related to the withdrawal of the wells, and the sample was considered to be representative of the total population in regard to ground-water withdrawals.
Metered withdrawal values for the wells in the test area were not available, so derived withdrawals were obtained by using power-conversion coefficients to convert power records to pumpage. The power-conversion coefficient is the ratio of values obtained in short simultaneous measurements of power use and discharge for a well.
Trend-surface and multiple-regression analyses were performed on the power-conversion coefficients themselves and on the pumpage per irrigated acre derived from the coefficients. The equations developed using the available data proved to be poor tools for estimating withdrawals.
In the analyses, all the expected relationships appeared to be either very weak or non-existent. The expected relations may actually be weak; however, a more likely conclusion is drawn that the withdawal values, which have been derived using the power-conversion coefficients and the available data, do not reflect the true withdrawals.
