ESTIMATING LOW‐FLOW QUANTILES FROM DRAINAGE‐BASIN CHARACTERISTICS IN NEW HAMPSHIRE AND VERMONT1

Authors:
  • SDS. Lawrence. Dingman
    S. Lawrence. Dingman

Abstract
This study systematically develops, validates, and compares alternative approaches for estimating quantiles of the distribution of annual minimum seven‐day‐average flows (7Q) for ungaged, unregulated drainage basins in New Hampshire and Vermont via regression on map‐measurable drainage‐basin characteristics. At 47 gaging stations in the region, the hypotheses that 7Q is log normally distributed and serially independent are not rejected, and the regional average spatial correlation is R= 0.35. Step‐forward examination of a suite of potential predictor variables revealed that logarithm of drainage area, mean elevation, and fraction of basin covered with sand and gravel deposits are significant predictors of quantiles of 7Q. The regression equations were incorporated into four approaches to estimate the 7Q value with a nonexceedence probability of 0.1, 7Q10. Comparison of observed values and values predicted via a delete‐one jackknife resampling validation indicates that one of the approaches gives estimates with acceptable bias and precision, with median relative error of 33 percent and prediction error of 64 percent. This is equivalent to the precision obtainable with only one to two years of gaging records. In spite of this limited precision, the approaches developed herein are useful for predicting 7Q quantiles at ungaged sites. Further improvement in precision will likely be possible only by exploiting the spatial correlation of annual 7Q.

This publication has 16 references indexed in Scilit: