HR: 1340h
AN: H33B-1388    [Abstracts]
TI: A comparison of regional regression and baseflow correlation for estimating low streamflow statistics
AU: * Kroll, C
EM: cnkroll@esf.edu
AF: SUNY College of Environmental Science and Forestry (ESF), 312 Bray Hall 1 Forestry Drive, Syracuse, NY 13210 United States
AU: Zhang, Z
EM: zzx509@yahoo.com
AF: SUNY College of Environmental Science and Forestry (ESF), 312 Bray Hall 1 Forestry Drive, Syracuse, NY 13210 United States
AU: Hirabayashi, S
EM: shirabay@syr.edu
AF: SUNY College of Environmental Science and Forestry (ESF), 312 Bray Hall 1 Forestry Drive, Syracuse, NY 13210 United States
AB: Estimates of low streamflow statistics are required for a variety of water quality and water quantity management purposes. While a frequency analysis may be performed when a historic record of streamflow discharges is available at the site of interest, a far more vexing problem is how best to estimate low streamflow statistics at ungauged river sites. Two statistical methods are often employed to estimate low streamflow statistics at ungauged river sites: regional regression and baseflow correlation. Here a comparison of these two methods is performed within a study region encompassing eastern Tennessee and western North Carolina, an area where lowflow regional regression models have previously performed poorly. New advances in both methodologies are explored with a goal of estimating the 7-day, 10-year low streamflow, a common lowflow design statistic. One reason lowflow regional regression models have sometimes performed poorly in practice is because important explanatory variables have not been available at ungauged river sites. Here we explore the use of new GIS-derived explanatory variables, including those from newly available maps of the baseflow index, time series of precipitation and temperature, as well as MODIS reflectance data during lowflow periods. In addition, using baseflow correlation we explore numerous methods for transferring information from multiple gauged sites to the ungauged site of interest. A jackknife simulation is developed to examine and compare the performance of the different estimation techniques. Initial results indicate that using multiple gauged sites improves the baseflow correlation method, especially when only a small number of baseflow measurements are taken at the ungauged river site. We investigate the tradeoffs between regional regression and baseflow correlation, and develop a scientifically based recommendation for the best method to estimate lowflow statistics at ungauged river sites within the study region. The experiment is also being developed in other hydrologic regions within the United States.
DE: 1812 Drought
DE: 1816 Estimation and forecasting
DE: 1869 Stochastic hydrology
DE: 1874 Ungaged basins
SC: Hydrology [H]
MN: Fall Meeting 2005