Robert B. McMaster, Eric Sheppard, Helga Leitner, Hongguo Tian, and Jeffrey Matson

In looking at the relationship between geographic information science/systems (GIS) and society, many, or even most, of the key issues--access, democratization, privacy--play themselves out in the area of environmental risk assessment. In particular, the application of GISs--and spatial methodologies in general--in order to better understand a given population’s exposure to technological hazards has increased rapidly over the past decade. One can point to a plethora of studies that have applied GISs to these human-produced hazards at a variety of scales and resolutions, and have applied significantly different spatial methodologies. Unfortunately, the effects of scale, resolution, and methodology are poorly understood. Additionally, the place itself, as constructed through a complicated set of economic, political, social, and economic forces and relationships, has rarely been used to develop a deeper and historical sense of the intermingling of the hazards and communities/neighborhoods in which they exist. Our multiyear project is attempting to unravel some of these vexing problems.

First we provide the context and definitions needed for our research. Since human environment hazards effect the air, water, and soil, this project, based in the Twin Cities metropolitan region, utilizes a series of hazardous materials sites, including TRI (Toxic Release Inventory), Superfund, Petrofund, and Land Recycling. However, most of the methodological work involves the airborne toxic releases produced by the TRI sites. In order to assess potential exposure to such sites (remembering that while exposure can be roughly calculated, risk itself IS based on a complicated set of physiological, historical, and toxicological variables) a set of both geodemographic (based on census data) and institutional populations were used. Whereas environmental risk assessment is the more comprehensive term that applies to assessing the relationships between all types of environmental hazards and the humans impacted, environmental justice or injustice is the special case of assessing the unequal spatial distribution of hazardous materials sites, where the assumption is that disadvantaged populations (the poor, minorities, the elderly) are disproportionately affected. In our own work, these hazardous materials and population data have been analyzed at a variety of scales and resolutions, ranging from the regional seven-county level to the individual neighborhood level, using a variety of spatial methodologies. This paper focuses specifically on our findings related to scale, resolution, methodology, and place.

Geographic scale is the extent of areal coverage used in an inquiry, and is applied in the opposite sense of cartographic scale, which is based on a strict mathematical relationship between map space and earth space using the representative fraction. Geographically, large scale represents larger areas; cartographically, large scale represents smaller areas (but with more detail). The misuse of these two terms causes significant confusion among the many interdisciplinary researchers that work in this area. Our work has looked at environmental justice issues at regional scales--the seven-county metropolitan area--to the individual neighborhood scale. Interpretations of environmental justice change as one looks at the problem from these differing scales.

The resolution, or granularity, of the data also affects measures of environmental injustice. Our data sets have applied census data at the census tract, block-group, and block levels. We currently are working with parcel-level data. Interpretations can change significantly when institutional data are included, such as that on schools and day cares. Further improvements will involve the addition of the 2000 census and approximations of day-time populations.

Whereas most studies use relatively simple measures for establishing the relationship between hazardous materials and at-risk populations, it is clear that more robust methodologies are needed. For instance, relationships are often based on simplistic measures of proximity--hazardous materials sites falling within a block group with a high percentage of minorities--which can not deal with differential toxicity, actual distance from site, and exact meteorological conditions. While GIS-based buffer analysis provides a better measure of proximity, it is nonetheless problematic in that all buffers are normally generated at the same size. The generation of plumes ameliorates this problem, but assumes that exact meteorological conditions are known ahead of time. Additionally, one needs to think carefully about the statistical significance of the existing distribution. To address this concern we have ventured into the realm of Monte Carlo simulation, where, through a randomization process, we can compute the sampling distribution, with respect to which the importance of observed statistics can be assessed.

THE neighborhoods in Minneapolis are intricate places with complex social and industrial histories. Questions of environmental risk assessment, and justice, have to carefully account for the historical sequencing of industrial toxic sites and migrations of populations in to and out of the neighborhood. This type of work involves a careful historical reconstruction using company records, census data, and historical land records, maps and atlases. In the end, claims of environmental injustice must include a meticulous accounting of the place, not just a modern analysis of easy-to-obtain institutional data. Additionally, we have discovered that the standard institutional databases such as TRI are often insufficient for characterizing a neighborhood’s environmental concerns, where many toxic sites must be determined through the acquisition of local neighborhood knowledge.

This study attempts to tackle multiple problems/issues related to environmental risk assessment, in particular as applied to the assessment of environmental justice. Using the Twin Cities region, we have systematically begun to better understand how the scales and resolutions used, the methodologies applied, and a clearer interpretation of place, can provide an improved understanding of this intrinsically geographical problem.