LANDSCAPE CHARACTERIZATION OF PERIDOMESTIC RISK FOR LYME-DISEASE USING SATELLITE IMAGERY

Remotely sensed characterizations of landscape composition were evalua ted for Lyme disease exposure risk on 337 residential properties in tw o communities of suburban Westchester County, New York. Properties wer e categorized as no, low, or high risk based on seasonally adjusted de nsities of Ixodes scapularis nymphs, determined by drag sampling durin g June and July 1990. Spectral indices based on Landsat Thematic Mappe r data provided relative measures of vegetation structure and moisture (wetness), as well as vegetation abundance (greenness). A geographic information system (GIS) was used to spatially quantify and relate the remotely sensed landscape variables to risk category. A comparison of the two communities showed that Chappaqua, which had more high-risk p roperties (P < 0.001), was significantly greener and wetter than Armon k (P < 0.001). Furthermore, within Chappaqua, highrisk properties were significantly greener and wetter than lower-risk properties in this c ommunity (P < 0.01). The high-risk properties appeared to contain a gr eater proportion of broadleaf trees, while lower-risk properties were interpreted as having a greater proportion of nonvegetative cover and/ or open lawn. The ability to distinguish these fine scale differences among communities and individual properties illustrates the efficiency of a remote sensing/GIS-based approach for identifying peridomestic r isk of Lyme disease over large geographic areas.