Regional Regression Equations to Estimate Flow-Duration Statistics at Ungaged Stream Sites in Connecticut

Abstract

Multiple linear regression equations for determining flow-duration statistics were developed to estimate select flow exceedances ranging from 25-to 99-percent for six "bioperiods"-Salmonid Spawning (November), Overwinter (December-February), Habitat Forming (March-April), Clupeid Spawning (May), Resident Spawning (June), and Rearing and Growth (July-October)-in Connecticut. Regression equations also were developed to estimate the 25-and 99-percent flow exceedances without reference to a bioperiod. In total, 32 equations were developed. The predictive equations were based on regression analyses relating flow statistics from streamgages to GIS-determined basin and climatic characteristics for the drainage areas of those streamgages. Thirty-nine streamgages (and an additional 6 short-term streamgages and 28 partial-record sites for the non-bioperiod 99-percent exceedance) in Connecticut and adjacent areas of neighboring States were used in the regression analysis. Weighted least squares regression analysis was used to determine the predictive equations; weights were assigned based on record length. The basin characteristics-drainage area, percentage of area with coarse-grained stratified deposits, percentage of area with wetlands, mean monthly precipitation (November), mean seasonal precipitation (December, January, and February), and mean basin elevation-are used as explanatory variables in the equations. Standard errors of estimate of the 32 equations ranged from 10.7 to 156 percent with medians of 19.2 and 55.4 percent to predict the 25-and 99-percent exceedances, respectively. Regression equations to estimate high and median flows (25-to 75-percent exceedances) are better predictors (smaller variability of the residual values around the regression line) than the equations to estimate low flows (less than 75-percent exceedance). The Habitat Forming (March-April) bioperiod had the smallest standard errors of estimate, ranging from 10.7 to 20.9 percent. In contrast, the Rearing and Growth (July-October) bioperiod had the largest standard errors, ranging from 30.9 to 156 percent. The adjusted coefficient of determination of the equations ranged from 77.5 to 99.4 percent with medians of 98.5 and 90.6 percent to predict the 25-and 99-percent exceedances, respectively. Descriptive information on the streamgages used in the regression, measured basin and climatic characteristics, and estimated flow-duration statistics are provided in this report. Flow-duration statistics and the 32 regression equations for estimating flow-duration statistics in Connecticut are stored on the U.S. Geological Survey World Wide Web application "StreamStats" (http://water.usgs.gov/osw/streamstats/index.html). The regression equations developed in this report can be used to produce unbiased estimates of select flow exceedances statewide. Habitat Forming March-April High Maintain natural habitat and connectivity with flood plain, channel formation, and flushing and transport of fine-grained sediment. Clupeid Spawning May Medium Increased flows...

Statistical Models to Predict and Assess Spatial and Temporal Low‐Flow Variability in New England Rivers and Streams

“…It is possible that I was able to discriminate among multiple effects due to the restricted time period of my analyses (1985–2015), whereas earlier models have used all available data. Predictive models for summer/July/August median flows and/or annual 7Q10 values have been developed by USGS for five of the six states in New England (Ries and Friesz ; Flynn ; Dudley , ; Ahearn ; Bent et al. ), but no models have been specifically developed for baseflows during summer months (New Hampshire models also include observations from Vermont, Maine, and Massachusetts).…”

Methods for estimating selected low-flow frequency statistics and harmonic mean flows for streams in Iowa

“…The Efroymson method is an automatic procedure for regression model selection when there are a large number of potential explanatory variables. The procedure is similar to forward selection, which tests basin characteristics one by one and identifies those that are statistically significant, except as each new basin characteristic is identified as being significant, partial correlations are checked to see if any previously identified variables can be deleted (Ahearn, 2010). When basin characteristics were found to be highly correlated to each other, only one basin characteristic at a time was tested in the Efroymson selection process.…”

Methods for estimating annual exceedance-probability discharges and largest recorded floods for unregulated streams in rural Missouri

“…To identify basin characteristics that are statistically significant for inclusion in the regression analyses, the Efroymson stepwise-selection method (Efroymson, 1960) was used to define potential explanatory variables from the list of 35 characteristics. The procedure is similar to forward selection, which tests basin characteristics one by one and identifies those that are statistically significant; however, as each different basin characteristic is identified as being significant, partial correlations are checked to see if any previously identified variables can be deleted (Ahearn, 2010). Highly correlated characteristics were included in the Efroymson selection method one at a time to avoid problems with multicollinearity.…”