Analysis of Select Stream Discharge Models in Southeast Alaska - Terry Schwarz , (co-authors: none)

ABSTRACT

Analysis of error was undertaken on four regionally-based statistical streamflow models in Southeast Alaska for the purpose of determining the overall fitness of the models when compared to current USGS gage records. Modern and historic data were utilized in a Geographic Information System (GIS) to extract basin characteristics for use as model variables. Spatial, temporal, and basin attribute trends were examined to establish if any factors showed an obvious relationship to relative model error. On average all models over-predicated discharge, with a mean relative error ranging from -0.1% to 20% and standard deviation of error ranging from 20% to 60%. No basin characteristics had strong correlation with model error. A cyclic trend was found in the error of the Mean Monthly Discharge (QAM) models, suggesting seasonal changes in modern streamflow statistics possibly due to climate change. No obvious trend in the spatial distribution of error was found, possibly because of the clustered nature of test gages.

Topic: forecasting and hydrologic analysis