A review and presentation on an article that documented a research effort that synthesized remote sensing and GIS techniques to assess the human risk from the Sin Nombre virus (SNV), the agent associated with hantavirus pulmonary syndrome. Deer mice, the primary rodent host, were studied in 144 field sites in the Walker River Basin in western Nevada and east-central California from 1995 to 1998. Spatial patterns and statistical relationships between site characteristics and infection rates were analyzed to retroactively classify the infection status of rodents in order to estimate prediction accuracy. These predictions could then be used to identify landscape characteristics where people are most likely to contract SNV.
The research experiment used Landsat Thematic Mapper to generate 100 ha unit maps of eight vegetation types, slope, vegetation density, hydrology and antibody data collected from deer mice. Two analyses were formulated: the vegetation approach was based on possible relationships between infection status and a pre-existing vegetation classification that may be relevant to SNV infections, and the discriminant function analysis (DFA) which generated a linear function that distinguished the properties of positive and negative sites. The DFA proved to be a more accurate analysis method.