In 1992, a remote sensing/GIS study was initiated at Ames to identify landscape features related to exposure risk for Lyme disease in Westchester County, New York. The study was funded by the Ames Director's Discretionary Fund. Epidemiological data and expertise were provided by investigators at the Medical Entomology Lab (MEL) of the New York Medical College and the Westchester County Department of Health. These data consisted of canine seroprevalence rates (CSR) for antibodies to the Lyme disease agent. CSR were estimated from blood samples taken in 1991 by county veterinarians, and represented, by municipality, the percentage of domestic dogs exposed to Lyme disease. The canine data provided a useful measure of the distribution of Lyme disease risk within the county. These data showed that risk appeared to increase from south-to-north along an urban-to-rural gradient. Landsat Thematic Mapper (TM) data acquired on May 20, 1991 were used to characterize land cover along this gradient. The land cover map and epidemiological data were integrated in a GIS to spatially relate landscape pattern and exposure risk.
An image processing technique known as an "unsupervised classification" was used to develop the land cover map from the TM data. Color-infrared aerial photography and field visits provided the information necessary to accurately identify ("label") these classes by cover type. The land cover classification identified different types of residential areas, as well as different kinds of vegetative cover important for ticks and their hosts. GIS functions were used to define individual forest patches. The area of each forest patch was then determined. Based on proximity to a forest patch, vegetated residential areas were then categorized as adjacent or not adjacent to forest. The land cover composition of each municipality was quantified by overlaying a coverage containing the muncipality boundaries, which also contained the CSR data, onto the new landscape map. As a result, a highly significant correlation was found between CSR and the proportion of vegetated residential area adjacent to woods (16K), as well as the proportion of forest, within a municipality. Overall, the study showed that basic relationships between the type and placement of landscape elements and Lyme disease risk at the municipality level could be described using an RS/GIS approach.
For more information about this research, contact Louisa Beck.