Estimating flood magnitude and frequency on streams and rivers in Connecticut, based on data through water year 2015

The U.S. Geological Survey, in cooperation with the Connecticut Department of Transportation, updated flood-frequency estimates with 50-, 20-, 10-, 4-, 2-, 1-, 0.5-, and 0.2-percent annual exceedance probabilities (2-, 5-, 10-, 25-, 50-, 100-, 200-, and 500-year recurrence intervals, respectively) for 141 streamgages in Connecticut and 11 streamgages in adjacent States using annual peak-flow data through water year 2015. Peak-flow regression equations were derived for estimating flows at ungaged stream sites with annual exceedance probabilities from 50 to 0.2 percent. Methods for estimating prediction intervals for the peak-flow regression equations are presented. The regression equations are applicable for basins in Connecticut with drainage areas ranging from 0.69 to 325 square miles that are not affected by flood-control regulation or flow diversions.

The flood discharges for select annual exceedance probabilities were estimated following new (2018) national guidelines for flood-frequency analyses. New guidelines have improved statistical methods for flood-frequency analysis including (1) the expected moments algorithm to help describe uncertainty in annual peak flows and to better represent missing and historical record and (2) the generalized multiple Grubbs-Beck test to screen out potentially influential low outliers and to better fit the upper end of the peak-flow distribution. Additionally, a new regional skew (0.37) derived for New England was used in the flood-frequency analysis for the streamgages.

Annual peak flows were analyzed for trends for four time periods (30, 50, 70, and 90 years) through 2015. Trend results show some statistical evidence of increasing peak flows in each of the time periods analyzed; however, multidecadal climate cycles may be influencing the number and magnitude of the trends. Historical peak-flow trends in and near Connecticut do not offer clear and convincing evidence for incorporating trends into flood-frequency analyses. For this study, the traditional assumption of stationarity is used with no adjustment for trends.

Generalized least squares regression techniques were used to develop the final set of multivariable regression equations for estimating flood discharges with 50-, 20-, 10-, 4-, 2-, 1-, 0.5-, and 0.2-percent annual exceedance probabilities. The standard error of prediction for the regional regression equations ranged from 26.3 to 45.0 percent. The standard error of prediction was slightly smaller in the current study compared to the 2004 study, indicating an improvement in the predictive ability of the equations (6 percent smaller at the 50-percent annual exceedance probability to about 1 percent smaller at the 1-percent annual exceedance probability). Generalized least squares regression techniques also were used to develop a one-variable (drainage-area-only) equation. Drainage-area-only equations can be used as an alternative to the multiexplanatory variable statewide regression equations if decreased accuracy is acceptable.

The revised statistical procedures and additional streamgage data applied in the current study result in a more accurate representation of peak-flow conditions in Connecticut than was previously available. The regional regression equations will be integrated in the U.S. Geological Survey StreamStats program, which estimates basin and climatic characteristics and streamflow statistics at user-selected ungaged stream sites.