Using GIS to Evaluate Pollutant Loads In the Root River Watershed

Primary Investigator: Dr. Vladimir Novotny

Sponsor: S.C. Johnson Fund Community Involvement Awards and Wisconsin

Foundation of Independent Colleges

Duration: September 2001 – May 2003

Funding Level: $10,000

The research goals of this project were:

(i) Quantify sediment load generated by agriculture and urban activities and delivered to the Root River

(ii) Compare present and future sediment loads

(iii) Identify crucial sediment generating areas.

ArcViewGIS, developed by ESRI, was utilized to perform calculations and display the results of the analysis.

Present objective results

Available data were collected over the past months from different agencies. Rainfall data have been obtained from the National Weather Service for Mitchell Field from 1948 till now, and three years were selected as representative for dry, wet, and normal years. Flow measurements were downloaded from the EPA BASINS database. Land use and soil maps in vector format were obtained from SEWRPC. After some consideration about the limited resolution of elevation data offered by EPA (300 m resolution), elevation data of 30 m resolution were downloaded from GeoComm International Corporation in a DEM format, and imported into ArcView in a grid format. A hydrologic extension was used to determine slope, trace the river system, and calculate the flow accumulation in the watershed.

The GIS model of diffuse pollution developed by the Institute (Blonn 2001) was based on the first version of the Universal Soil Loss Equation to determine the sediment load from rural areas. It also was implemented in a vector environment. For this analysis, the initial idea was to rely on the vector approach, but the decision to shift into a raster environment derived intrinsically from the model used to calculate the sediment load from rural areas: a revised version of the USLE (RUSLE3D) developed by Mitasova at the University of Illinois at Urbana-Champaign. The Model Builder extension offered by ArcView represents a useful tool that allows customers to develop spatial models inside ArcView GIS. In this study, it was used to automate calculations for the RUSLE3D (Figure 1).

Figure 1: Example of Model Builder Erosion model for 50% rainfall data (average years)

Present work and future goals

The load from rural areas currently is underestimated using the SCS method combined with Event Mean Concentration for pollutants of concern. A custom-developed script is necessary to deal with the great amount of layers originated from runoff calculation on a daily basis.

This happens because the surface runoff calculation is implemented in a raster environment and daily values are summed to obtain an annual-base final result. The three representative years for precipitation were chosen as suggested above.

All calculations are done in ArcView using extensions and developed scripts. Some of them still need to be developed. The latest goal is to assemble everything into an extension, so that, providing all necessary data, the same calculations can be automatically performed for similar problems in other watersheds.

The problem of sediment routing is under consideration and study. Up to now, a subwatershed-differentiated delivery ratio seems the best solution, but the hypothesis of using a water quality model to transfer the sediment load should be analyzed.

Water quality data were obtained from the Milwaukee Metropolitan Sewerage District (MMSD) at some stations in the upper part of the Root River. Unfortunately, at a first sight, the data do not seem sufficient for an accurate calibration of the model, but should be sufficient to check an estimate of the total load. Flow data also will be used to calibrate the model using the annual surface runoff determined by the main script developed.

Point sources of pollution will be obtained from the BASINS database and added to the non-point sources to get the total load of pollution into the river. Calibration for pollutants will be accomplished for the upper part of the watershed, and results will be extended to the whole watershed. More data available can be helpful.

Conclusions

ArcViewGIS is able to convey a spatial distribution of loads generated (for example, a distributed load from rural areas for the dry years is presented as a map in Figure 2).

Figure 2: Spatial distribution of soil loss (in tons/acre*year) for the 25% percent rainfall data (as representative for dry years).

< Institute for Urban Environmental Risk Management