UPDATES: July 2011

UPDATES is a monthly email newsletter on stormwater management, assessment (including monitoring), and maintenance research at St. Anthony Falls Laboratory and the University of Minnesota.

Click HERE to subscribe to UPDATES Newsletter

Or Text SWUPDATES to 22828 to Join. Message and data rates may apply.

For Email Marketing you can trust


Techniques for Determining Effective (Connected) Impervious Area

July 2011 (volume 6 - issue 6)

Contributed by Ben Janke and John S. Gulliver (Department of Civil Engineering, University of Minnesota)

Practitioners responsible for the design and implementation of stormwater management practices rely heavily on estimates of impervious area in a watershed. The most important parameter in determining actual urban runoff is the ‘effective’ impervious area (EIA), or the portion of total impervious area that is directly connected to the storm sewer system. EIA is often considerably less than total impervious area, and can vary with rainfall depth and intensity. Despite its importance, EIA is often not well known in practice. A more accurate estimate of EIA would aid the design of stormwater management practices by improving the accuracy of hydrologic simulations and providing a means to assess the impact of disconnection on discharge from a watershed. These outcomes should result in more effective and properly designed stormwater management practices.

Two primary approaches to modeling EIA include: (1) analysis of rainfall-runoff data in gauged watersheds, and (2) analysis of spatial data such as land cover and elevation. The latter approach is particularly appealing because rainfall and runoff data are not always available or of sufficient quality and resolution to be used for analysis. Furthermore, the coverage and quality of spatial data are constantly improving, allowing more accurate GIS-based methods to be developed.

One particular method, developed by Han and Burian (2009), estimates effective impervious area from automated analysis of fine-scale GIS data layers for land cover, elevation, and stormwater collection infrastructure (e.g. catch basins, open channels). The method determines EIA by tracing flow paths from ‘paved’ cells in the land cover grid, partitioning these cells into those that eventually drain into stormwater collection cells (‘connected’ impervious) and those that drain onto pervious surfaces (‘disconnected’ impervious). Roof connectivity can generally only be determined from site inspections, and is therefore left as a user-specified parameter by the tool.

The Capitol Region Watershed District, a 41-square-mile urban watershed comprised by a majority of St. Paul and parts of surrounding cities, was a suitable watershed for application of the GIS tool due to the availability fine-scale (1.0-m) elevation and land cover data (courtesy of M. Bauer, Department of Forestry at the University of Minnesota), and a layer for the impervious surfaces in the watershed (courtesy of Capitol Region Watershed District (CRWD)). CRWD also provided rainfall and runoff data for several sub-watersheds, which were used to provide a second estimate of EIA using the analysis technique of Boyd et al. (1994).

The tool was first applied to a small (40-acre) watershed near Como Lake that drains to an underground stormwater vault monitored by CRWD (Figure 1). This watershed is primarily old residential, with significant tree cover (29%) and alleys instead of driveways, thus EIA was expected to be relatively low. A first pass of the tool produced a connected pavement area of 8.0% (EIA of 8.0% assuming no roof connectivity.)

Figure 1: Land cover for test watershed near Como Lake in St. Paul (note shading of roads and alleys). Drainage areas of infiltration basins and rain gardens are shown, but are assumed to not contribute water to the outlet.

Next, the rainfall-runoff data were analyzed for the catchment using the method of Boyd et. al (1994). In this method, the runoff depth (runoff volume normalized by total watershed area) is plotted versus rainfall depth for every storm in the record. The slope of a regression line fit to this data is the percentage of total watershed area contributing water to the outlet. Boyd et al. (1994) recommend removing points lying further than 1mm of runoff depth from the line (which likely involve runoff from both pervious and impervious surfaces) and re-calculating the regression. This process is repeated until all points are within 1 mm of runoff depth of the line; the slope is then assumed to reflect the connected impervious area only. This process produces an EIA value of 14.8% for the test watershed (Figure 2), considerably higher than the first estimate of 8.0% from the GIS tool. Roughly 30% of the rooftop area would have to be connected to the drainage system to match the data, which seems unrealistically high for this watershed.

Figure 2: Plot of runoff depth vs. rainfall depth as observed at the monitoring site. 165 events occurred over the 2007-2010 monitoring periods. EIA is 14.8% (slope of regression line).

The discrepancy between the data analysis and GIS tool estimates of EIA is likely due to a limitation of the GIS tool for highly shaded watersheds. Tree canopy obscures a significant portion of paved area that is almost certainly connected to the drainage network (roads especially), with the result that the tool produces erroneous ‘pockets’ of disconnected road.

A second run of the GIS tool was conducted by modifying the land cover layer to ‘un-shade’ the roads and alleys. This modification produced a pavement connectivity of 14.9%, which matched the data-derived EIA (assuming zero rooftop connectivity; see Figure 3). Given that some rooftop connection is probably present, the actual road connectivity for the watershed is likely between the two estimates (8.0% to 14.9%).

Figure 3: Land cover map for test watershed following the second application of the GIS tool (i.e. with roads and alleys forced to be ‘un-shaded’). Connected pavement makes up 14.9% of total watershed area.

In future work, the GIS tool will be applied to larger and more diverse watersheds. A major limitation of the method is the dependence on a roof connectivity estimate; a better measure of rooftop connection for a watershed (e.g. from site surveys) would allow the EIA value to be estimated with greater accuracy. The rainfall-runoff analysis does not have this limitation, and tends to also be less time-consuming to implement, once the monitoring and rainfall have been established and are operated.

A preliminary conclusion is that rainfall-runoff data analysis will be the better method of determining connectedness (EIA) in a watershed. However, in un-gauged watersheds where such data does not exist, the Han and Burian (2009) GIS method has potential to allow practitioners to estimate EIA with reasonable accuracy.


  • Boyd, MJ, MC Bufill, and RM Knee. 1994. Predicting pervious and impervious storm runoff from urban drainage basins. Hydrological Sciences Journal, 39(4): 321-332.
  • Han, WS, and SJ Burian. 2009. Determining effective impervious area for urban hydrologic modeling. Journal of Hydrologic Engineering, 14(2): 111-120.


Citation: "Stormwater Research at St. Anthony Falls Laboratory." University of Minnesota, St. Anthony Falls Laboratory. Minneapolis, MN. http://stormwater.safl.umn.edu/