Runoff and Chemical Loading in Small Watersheds in the Twin Cities Metropolitan Area, Minnesota

Flow, rainfall, and water-quality data were collected during 1980 for 15 to 30 rainfall and snowmelt events on 6 rural and 11 urban watersheds in the Twin Cities Metropolitan Area. Event or daily flow and load models (for seven constituents) were developed and used with runoff and rainfall data for 1963-80 to compute 2-year frequency annual and seasonal flows and loads for each watershed. In models of storm-sewered watersheds, total storm rainfall proved to be the most significant factor controlling runoff and loads. Depending on the watershed type, antecedent soil-moisture indices or rainfall intensity also were important factors in estimating runoff. Annual runoff from storm-sewered watersheds averaged about 27 percent of annual precipitation, ranging from 13 to 57 percent. Runoff in urban main-stem streams ranged from 13 to 20 percent and was related to the percent of urbanization in the watershed. Annual runoff in rural watersheds ranged from 6 to 20 percent of annual precipitation. The percentage storm-runoff response increased with increasing storm size for watersheds with storm sewers, more so for the steeper than the flat watersheds. Runoff responses ranged from 2 to 22 percent of rainfall for a 0.1-inch rain and from 14 to 52 percent for a 2-inch rain. As much as 75 percent of the runoff from a 2-inch rain was derived from previous areas in the steeper, storm-sewered watersheds. As a result of storm and seasonal differences in runoff response to rainfall, the seasonal distribution of runoff and loads for the rural watersheds did not follow seasonal rainfall patterns. Instead, runoff and loads were greatest in the snowmelt period and declined through the year in response to decreasing soil moisture and pervious-area runoff. Seasonal runoff and loads from urban watersheds more closely followed seasonal rainfall patterns with a maximum in the summer. Urban storm-sewered watersheds responded to virtually every rainfall event. Based on 18 years of simulated record, the runoff and loads expected to be equaled or exceeded on a long-term average of every 2 years in each watershed were used in an attempt to generate regression models with basin characteristics. Unfortunately, all attempted groupings of sites yielded unreliable models. There is need for load-mitigating measures in the steeper, more urbanized watersheds of the metropolitan area. Wetlands and other low-intensity land uses were found to be important factors in controlling loading of urban and rural streams. Loads in rural streams were highly dependent upon factors affecting runoff. Therefore, practices that increase rainfall retention near the source or that reduce channel conveyance likely will reduce rural runoff and loads.