Overland runoff from urban areas can cause concentrations of some water-quality constituents in local receiving streams to increase. The U.S. Geological Survey in cooperation with the Kansas Department of Health and Environment investigated the water-quality characteristics of streams draining Topeka, Kansas, and adjacent parts of the Shunganunga Creek basin from October 1979 through November 1981. The purpose of this investigation was to provide the data and interpretation necessary to determine the effects of runoff from urban areas on the water-quality characteristics of receiving streams.

Water-quality characteristics for three streamflow conditions were determined: (1) dry-weather streamflow--a combination of base flow and point-source contributions, (2) storm streamflow--mainly provided by overland runoff from storms, and (3) snowmelt streamflow- mainly provided by overland runoff from snowmelt.

Median concentrations of trace metals and nutrients were larger in storm streamflow than in dry-weather streamflow. Median concentrations of total lead and zinc were largest in storm streamflow from the more urban basins. Regression equations were developed to estimate median concentrations of total lead and zinc in storm streamflow from the percentage of drainage area in residential plus commercial land-use areas (correlation coefficients were 0.98 for total lead and 0.88 for total zinc); and from street density in lane miles per square mile (correlation coefficients were 0.89 for total lead and 0.84 for total zinc). Median concentrations of dissolved nitrite plus nitrate nitrogen and total phosphorus averaged, respectively, 76-percent and 70-percent larger during storm streamflow than during dry-weather streamflow and were largest in storm streamflow from the more agricultural basins.

Median concentrations of dissolved sodium, chloride, and solids in snowmelt streamflow at all study sites averaged 218-percent larger for dissolved sodium, 296-percent larger for dissolved chloride, and 71-percent larger for dissolved solids relative to median concentrations in dry-weather streamflow. Regression equations also were developed to estimate median concentrations of dissolved sodium, chloride, and solids in snowmelt streamflow from the summation of percentages of the drainage area in residential, commercial, and industrial land-use areas (correlation coefficients were 0.97 for each of the three relationships) and from street density in lane miles per square mile (correlation coefficients were 0.93 for dissolved sodium and 0.94 for both dissolved chloride and dissolved solids).

Multiple-correlation and regression analysis relating storm-runoff volumes and average constituent concentrations to land-use and storm characteristics produced significant relations (0.05 level of significance) for storm-runoff volume (0.81 coefficient of determination), total lead (0.71 coefficient of determination), total zinc (0.50 coefficient of determination), and suspended sediment (0.58 coefficient of determination).