Climatic and physiographic drivers of peak flows in watersheds in the North Slope of Alaska

Abstract

The failure to accurately predict peak discharge can cause large errors in risk analysis that may lead to damage to structures and in some cases, death. Creating linear regression (LR) equations that accurately predict peak discharges without historic data provides a method to estimate flood peaks in ungauged watersheds on the North Slope of Alaska. This thesis looks at the independent variables that drive, or are significant in predicting snowmelt peak discharge in the North Slope watersheds. The LR equations created use independent variables from meteorological data and physiographic data collected from four watersheds, Putuligayuk River, Upper Kuparuk River, Imnavait Creek and Roche Moutonnée Creek. Meteorological data include snow water equivalent (SWE), total precipitation, rainfall, storage, length of melt. Physiographic data summarize watershed area (2.2 km2 to 471 km2) and slope (0.15:100 to 2.7:100). This thesis compared various Flood Frequency Analysis techniques, starting with Bulletin 17B, multiple USGS regional methods and finally created LR equations for each watershed as well as all four watersheds combined. Five LR equations were created, three of the LR equations found SWE to be a significant predictor of peak flows. The first equation to estimate peak flows for all watersheds used only area and had a high R2 value of 0.72. The second equation for all watersheds included area and a meteorological independent variable, SWE. While the evidence presented here is quite promising that meteorological and physiographic data can be useful in estimating peak flows in ungauged Arctic watersheds, the limitations of using only four watersheds to determine the equations call for further testing and verification. More validation studies will be needed to demonstrate that viable equations may be applied to all watersheds on the North Slope of Alaska.