Brian Altonen, MPH, MS

NOTICE!!

A Survey has been developed to document and compare GIS utilization in the workplace.  This survey assesses GIS availability and utilization in both academic and non-academic work settings.  The purpose is to document the need for GIS experience as an occupational skill.   GIS is currently being underutilized by most companies.  Spatial Technician and Analyst activities and a few managerial activities requiring GIS are reviewed. 

 This survey, which takes about 20-25 mins to complete (’tis a bit long),  can be accessed at

 Survey Link

……………………………………………………………………………………………………

[Note: The full report filed for the grant-funded portion of this work, prior to developing many of these techniques, appears on this site at: https://brianaltonenmph.com/3-gis-environmental-health/report-for-grant-funded-research-2002/]

.

GIS Spatial Analysis of Oregon Toxic Release Sites

The primary purpose for mapping toxic release sites is so we can develop a better understanding of how these sites spatially may relate to the various illnesses they are purportedly related to. In the public eye, and the eyes of the popular press, these sites are somehow very much related to the various illnesses a local community develops at time. Toxic release sites have been related to various forms of cancer, birth defects, mental illness, learning deficiencies, ADA, losses in vision, skin-related problems, migraine headaches, epilepsy, lymphoma, asthma and COPD . . . the list goes on. Trying to link a chemical spills site to a specific malady is very much a hit or misss operation, with outcomes that often appear more successful in the public eye than in the science or mathematical world–the basis for much of the GIS used to scientifically try to link chemical releases to specific forms of illness, especially cancer. We can use GIS to first explore the spatial relationships between chemical spills and cancer visually and spatially, and then use this information to either develop new ways to explore GIS applications to monitoring chemical releases, or take the statistical route in evaluating the mapped data.

In GIS, there are numerous ways to calculate spatial relations between chemcial release sites and medical conditions. The fact is how ever that it becomes close to impossible to demonstrate strong statistical relations with disease that have a fairly rare occurance. An even if we do come up with an outcome with our numeric evaluation that meets the needs for being noticeable in a statistical sense, the statistical significance attached to this outcome and any claims made in the end using these findings, just take advantage of results that may have happened simply due to probability. It ends up, that if we explore enough types of relationships over time, there is the likelihood that a positive finding will surface that appears statistically significant. But if the numbers of trials made before reaching this success appear plenty, that outcome in your favor as a researcher should be interpreted as being simply the result of the probability equations for these outcomes (a 95% probability of significance means a 5% probability of failure, which several hundred tests later of various mapped areas, may not be signficant). If you really want to prove that your finding is valid, you need to obtain the same outcome and conclusion using another area that is comparable spatially, with many of the same spatial and demographic relationships. Otherwise, you are left with just a possible finding that is simply in itself a non-repetitive, non-confirmable outcome. C’est la vie with chemical and cancer mapping and GIS.

In spite of this doubtfulness I express with GIS and cancer mapping, I still find it to be quite useful from the public health perspective as a monitoring tool. For should the time come when more repeatable, and proveable outcomes do surface spatially, we want to be ready to better udnerstand and prevent these problems from erupting further or spreading to other areas. GIS is very a helpful tool for predicting disease ecology-related features, if we view the use of GIS as the means to monitor environmental chemistry and disease, we again are emplying this GIS in a more preventive fashion and educational fashion, without all the bells and whistles of systems used to prevent some disease from spreading in a fairly quick manner.

For this review of Oregon chemical release sites, all datasets are equally applicable to these projects, but the focus tends to be on several fairly toxic parts of the state, at least in theory. Most of the work has focused on the Portland area, but some of it explores ways to view toxicology data pertaining to other urban settings in the state such as Eugene, Astoria, Coos Bay, and the various shipping towns along the Columbia River.

A number of methods are used to explore these areas spatially, starting with standard grid analysis and buffer techniques, and ending with the use of a hexagonal grid mapping method I developed to better describe impacts as a spatial feature, and to make more accurate products in these attempts to map the distribution of chemical risk across the state.

Portland, Oregon Metropolitan Area Investigations

Development of the Study Area for Exposure to Chemical Release in a Metropolitan Area

Defining the metropolitan study area - the use of a 30 mile band. (Red points/shapes depict superfund, superfund applicant, high risk and CRI sites)

A significant amount of time was spent evaluating the Portland, Oregon metropolitan area and neighboring suburban and tri-city regions. The primary reason for this emphasis on the Portland area was the case clustering noted for several cancer types related to toxic chemical release in this area, and the extremely high number of chemical release and spill reports filed for this part of the state. The Portland area is also of interest due to its proximity to a major waterway, the large numbers of unique chemicals industries found in this part of the state, and the large amount of public and political pressure and support related to the possibility for a significant or disastrous form of chemical exposure in such an urban setting.

In urban settings like Portland, the clustering of cases is usually considered to be largely due to people behaviors, activities that are not related to exposure- or carcinogenic-related events. These activities engaged in by the public include removing to areas where you can receive more effective care for you problem in the larger medical systems usually set up in these settings. The likelihood of obtaining significant amounts of the most up-to-date tertiary care, more rapidly, by a specialized health care service, usually makes moving to such areas an important step to be taken by anyone seeking a cure. Along with this tendency for people to populate and aggregate in certain areas (ill or not), we see increased risks of exposure to chemicals other than those produced by business, such as the automobile exhaust of cars and trucks, the need to obtain water made potable by various agencies, to the need for changes in air quality assessment methods due to increased density of people who smokes, rely upon an air condition, or make use of potentially toxic paints and the like. In addition, the increased need for available public transportation in areas where cancer treatment is found might also be a related reason for the increased risk of residing in such settings. For whatever reason, the fact that about half of the state’s cases are in this urban settings, along with mroe than half of its various types of toxic release sites (superfund, superfund applciant, confirmed release, etc.) suggest that at least an introductory exploration of the chemical history of the Portland metropolitan region needs to be done as a part of this project.

The second reason the Metropolitan region was focused upon was the high density of toxic release sites, including superfund and superfund applicant sites. Fifteen 15 out of 22 highly toxic sites identified by my studies as sites high risks sites were not considered applicable for superfund coverage around 2000. Reviewing the history of these sites it became clear that these sites needed to be reviewed as if they were superfund sites, mainly due to their close primity to the core of the Oregon metropolitan setting and their close proximity to some of the poorest census blocks within this urban setting.

Soon after this study began one additional reason surface for the review of these chemical sites. One of the largest such sites in the state — the Multnomah groundwater site — was finally granted superfund status. At the time, I had just completed two detailed GIS research projects on the history of release of pollutants from two fairly large industries–a metal manufacturing facility with almost fifty years of continuing manufacturing and release history, and a paper mill with more than 75 years of chemical release history. I had just explored the concept that the pollutant from these two major release sites overlapped in the area close to the Multnomah groundwater site. I already had a good idea on the topography of this area and its relationship to industrial waste behaviors in the local air and water spaces. Using this as a jump-off point for my research, I switched from Idrisi to ArcView-Spatial Analyst mapping and so began my Portland area superfund-superfund applicants-high risk site chemcial release project.

To define the Portland area, the analysis began with the definition of this region based on the work of the local metropolitan agency responsible for overseeing and mapping all regional planning activities. The local metropolitan group had already defined the urban boundary for the Portland Area. Although this boundary tended to change from year to year, its overall distribution seemed the change very little. In part because there were well defined political boundaries already existing in the outskirts of the city, defining regions which already had significant political supervision by non-urban agencies. Population density differences were also the reason for the fairly limited growth of the urban boundary at the time. Once you left the urban boundaires, in most directions, you headed into what was obviously non-urban living areas, where topography payed a key role in determining whether or not the remaining lands could be considered part of the urban boundary, at least for the time being.

So, I felt I had the urban boundary pretty well defined for this study of population density and features in relation to chemical release site lcoations and environmental conditions. This urban boundary was then smoothened out to produce an ovoid shape later so I could apply it to my palns to produce a Theissen’s polygon evaluation of the metro region in order to demonstrate where the highest risk areas were located in the three counties forming this urban setting.

To better develop our initial understanding of the spatial distribution of chemicals releases in the Portland region, I developed centroids for each of the four parts of the Portland area. The Northwest sector, where the majority of chemical release sites were lcoated, formed the high risk region in the Portland area. Three other sites defined as sectors were also defined, to distribute the remaining urban area land are surface and population features fairly equally across the remaining urban region. Each of these sectors had to have one or more fairly significant toxic release to which the local population would be exposed. I then evaluated each of these regions in terms of toxicity, carcinogenicity, types and degrees of potential exposure, and how these superfund, applicant and high risk release sites related to local cancer cases.

An overview of the Portland study area - population features

At first I thought it would work best to compare one set of regions clsoe to release sites to regions distanced from release sites. So long as these sites had similar population features, this type of study might do for the time being. This resulted in the following plans to evaluate chemical sites to cancer cases. Because this method of analysis is not a typical method employed for such studies, I was slow to take it on, and ultimately, it never had to be performed due to other methods for analysis recommended by other experts in the field.

The next method used in this analysis was the standard Monte Carlo evaluation, in which probabilities for disease clustering are analyzed spatially by looking at local population density details, and then determining whether or not case counts (case density or incidence) seemed higher than the model tested for by a computer using known standard incidence and disease count rates. This testing of the distribution of the cases unfortunately resulted in a map pointing to an area with both fairly low population density and low case counts, and fairly low chemical release site counts, so low in fact that it would be hard to explain these outlying high risk cases using the standard spatial methods this research was developed to test and to employ.

In the end, this led me back to the Theissen polygon tool use for analyzing chemicals first, then cases. For the Portland area, the following map was produced following the first use of the Theissen Polygon ArcView Avenue Extensions tool. Now it was apparent, that there was much more we could do to compare the four sectors of Portland in relation to cases and chemical release site history.

Chemical Release Site Density and Exposure Risk. The Northwest Portland Metropolitan Area

When reviewing chemical release sites, there are old sites that have been cleaned up, old sites that have not been cleaned up but have little to no risk of causing carcinogenesis, recent sites that have been cleaned up and are therefore no longer carcinogenic if they originally were, and recent sites that have not been cleaned up that may or may not be carcinogenic depending on their site chemistry. The severity of the more recent and potentially toxic sites is told to us by the status of the site: is it a Superfund site? is it a Superfund applicant? does it have an exceptionally high toxic chemical release history but is still not included in the Superfund cleanup or application process. On several occasions, there were sites identified in the Portland area that should have received immediately approval for Superfund clean-up, but for whatever reason(s) never qualified for this status.

Site Distribution in relation to high risk census block group areas

The various types of Portland Toxic Release Sites. All sites historical and recent, toxic or not.

The above map depicts all reported chemical release sites, toxic or not. Many of these sites were originally reported, but never had chemical reports filed due to insufficient evidence, improper reporting procedures, hearsay-like reporting procedures, and lack of credibility of the report following an initial investigation. These numbers do not include reports that were never reviewed for whatever reason, to which no address or reporting information could be found in the database used to store chemical release sites information. These sites are those found in the active database on Oregon chemical release sites, and those of the list of historical sites considered cleaned-up, this list found elsewhere on the web back in 2003.

Confirmed Release Inventory (CRI), High Risk, Superfund Applicant and Superfund Sites

The remaining sites once non-toxic sites and other low risk sites are removed.

When sites that are considered non-low risk sites (remediated or not) are removed. we are left with some Toxic Release Sites with well-documented chemical history and amounts of corresponding chemical release, Confirmed Release Inventory sites (CRIs, which have proven and field testing done on their chemical histories), High Risk site (HR, as defined by this researcher, these requirements noted later), Superfund Applicants (SFAs, based on a federal listing of superfund applicants sites developed before the actual superfunds were selected during the late 1980s), and Superfund sites (SF, those established by around 2003, the year this study went into full speed).

TRIs, CRIs, SFAs, SFs, HRs

The dataset developed for this study also bore historical data. This was used to identify high risk sites based upon length of time the chemical releases may have taken place.

DEM modeling of the high risk area– Portland to Troutdale–in relation to local toxic release history and watershed chemistry. [no image identified]