STREAMFLOW STATISTICS

DATA

Streamflow gaging stations whose records were used in the study are included in the Hydro-Climatic Data Network of the U.S. Geological Survey (Landwehr and Slack, 1992). The HCDN data base contains streamflow data for 1,659 gaging stations. The streamflow analysis in the Atlas was restricted to stations in the 48 contiguous states and excluded stations with less than 20 years of data, basins with drainage area greater than 50,000 sq. mi., and basins with noncontributing area greater than zero. There remained 1,456 stations, with an average record length of 45 years.

The HCDN file BASINS.DAT reports basins characteristics for HCDN stations. Unfortunately the file is incomplete, contains many suspect values, uses some ill-defined basin characteristics, and appears not to have been updated since the 1960's. It was held to be unusable as a basis for statistical analysis.

ANALYSIS

Regionalization of stream gaging stations is inherently more difficult than is the case for precipitation stations, because of the complications of geology, land use, streamflow diversions, etc. The stations in the HCDN do not have reliable information on watershed characteristics, and a physically reasonable grouping of gaging stations into homogeneous regions could not be achieved. Consequently, at-site distributions were calculated for individual stations in the HCDN. While this is less satisfactory than a successful regionalization, it does enable the analyst to easily obtain the quantiles for stream gaging stations, and should materially reduce the amount of work required to undertake a drought analysis of streamflow for durations of one month or longer.

It was not thought profitable to try to identify frequency distributions for streamflow separately for each gaging station. Accordingly, a single distribution was used for each duration at each site. The Wakeby distribution was chosen, because it can assume shapes typical of a wide range of plausible streamflow distributions. The Wakeby distribution was fitted to the L- moments of the at-site data, using Hosking's (1991) implementation of the algorithms of Landwehr et al. (1979). The longest duration for which frequency distributions was calculated is 12 months. It was considered that record lengths were too short to reliably calculate distributions for longer durations.

CONCLUSIONS

The high degree of uniformity exhibited for precipitation data is not evident for streamflow, even for the 12-month duration quantiles. Consequently, no conclusions concerning patterns in the data were drawn. The at-site quantiles are given in the streamflow tables.

Return to Main Page, National Drought Atlas

Precipitation Statistics, in "Clusters" selected from a U.S. map

Streamflow Statistics, by stations within USGS Hydrologic Regions, selected from a U.S. map.

revised 1 Aug 2006

Return to National Drought Atlas home

Precipitation Statistics, in "Clusters" selected from a U.S. map

Streamflow Statistics, by stations within USGS Hydrologic Regions, selected from a U.S. map.