The mission of this project is to identify opportunities for environmental improvement in the area of the Rouge River watershed. The method involves analyzing the spatial aspects of water quality in the watershed in order to identify targets for policy implementation.
The mission has changed somewhat since this project started. Initially, the objective was to identify opportunities for greenway development, particularly around the Rouge River. Using fractals and self-similarity as an approach to locate areas of concern regarding drainage, was a second step. The northern area of Detroit, for example, does not show any drainage (see flood-prone areas). Pre-settlement maps, however, show that there were many intermittent water bodies in the area, which were probably paved over. Finally, this study focused on water quality in the Rouge river.
The River Rouge has been studied extensively, and sampling data on various contaminants can be found at the River Rouge Project website. The way it is set up, it is difficult to see how water quality regarding specific contaminants varies spatially. In an initial attempt to show spatial variation of water quality, I digitized the sampling points and converted all the data tables into one single table with the average concentration of the main contaminants sampled at each point, and mapped the data with graduating symbols.
It is reasonable to expect that water quality is especially deteriorated around the river's mouth in Detroit, given the industrial land use of that area. If we see them separately, we can see already that for some contaminants the critical points are not necessarily located within Detroit's industrial area. An animation of these maps allows to see this more clearly. Spatial visualization of this information is useful because watersheds are multi-jurisdictional regions, which are challenging to manage coordinately. Critical sites for specific contaminants can be identified, and therefore efforts can focus on eliminating the specific pollutants in each jurisdiction. This is very significant, considering the cost of control and prevention of water pollution. In order to study the spatial distribution of specific contaminants, more detailed data was obtained for a 5 year period (1994-1998), and statistical analysis (ANOVA, discriminant analysis and principal components analysis) were perfomed to identify relationships between water quality and land use. Land use patterns in the areas of major concern might indicate important sources of pollution and target areas for policy action.
Water Quality-Land Use Relationships
Detailed data on water quality for the Rouge River Basin was obtained from the Wayne County Rouge River National Wet Weather Demonstration Project (1999). This data was processed, in order to determine the variation of concentration of specific pollutants over the time period between 1994 and 1998. Not all sites were sampled every year for all contaminants, making any comparative analysis difficult. However, animation maps allow to see that changes in concentration occur through time, and the variation pattern might indicate the response to specific contaminating events, rather than due to more permanent land use patterns. The animated maps of those contaminants that were sampled more frequently are listed below:
Land use and percentage of impervious surfaces are claimed to affect the water quality of surface water bodies. In the Rouge River Basin, land use information was obtained from SEMCOG (1995) and compressed into fewer categories, depending on estimated effect on water quality. A denser or more intensive use of land can be seen towards the City of Detroit, while a greater area dedicated to agricultural use seems to prevail towards the west (Washtenaw County) and the north of Oakland County. Impervious surface is related to land use (SEMCOG, 1995), and the highest percentages occur in the direction of the City of Detroit.
A specific land use and a percentage of impervious surface were assigned to each sampling site, using the Intersecting Themes feature of ArcView. Given the limitations of the contaminant data, average data for 21 sampling sites in the period 1994-1997 was used to conduct the statistical analysis. An ANOVA (analysis of variance) tested for significant differences between the water quality defined by the contaminants listed above, at sites located in different land use and percentages of impervious surface.The results show that there was no significant relationship.
A discriminant analysis showed that there is a 30% overlap between the groups defined by sensitivity to imperviousness. This might be a good result, but corresponding eigenvalues were too low and Wilkin's Lambda values were too high. The discriminant analysis using land use categories, on the other hand, resulted in much higher eigenvalues and lower Lambdas. However, the error was of 35%. Some grouping may be observed for less intensive use of land (see graph). Comparing built with non-built environments, the results of the discriminant analysis have a lower error, 25% (see graphs). However, the eigenvalues and Lambdas are not very good.
Finally, the principal components analysis resulted in a considerable overlap by imperviousness factor, using the first three components that explain most of the variation. Nevertheless, some grouping can be observed by built factor especially, and also by the urbanized factor (see graph). Built environment refers to either urbanized or non-urbanized areas that are densely built and have various service infrastructure. An urban park, for example, would be an urbanized, but unbuilt area.
The main contribution of this type of analysis is the spatio-temporal visualization of water quality. A better analysis would be possible with data from systematic sampling of a larger amount of sites through time, and a history of important discharge events to the waterways. This might allow to identify more significant relationships between water quality and land use.
Even though the resuts are not conclusive, there seems to be a closer
relationship between water quality and land use, than between water quality
and impervious surfaces. In particular, low intensities of use seem to
have lower impacts, even if the activity corresponds to an open area used
for agricultural activity. This supports the idea that not only the amount
of open areas will impact the water quality in a river basin, but also
the type of activity performed in such open area. In other words, reducing
impervious surfaces alone would not have significan impacts. Forested wetlands
and other types of wild ecosystems, together with low impact development,
should also be the aim of restorative efforts in the Rouge River Basin.
In the City of Detroit, there are plenty of opportunities for this kind
of effort, considering the amount of vacant land that the city can develop
into green areas closely associated to the Rouge River Basin.
Maps and Sources of Data