Abstract
Tick-borne encephalitis (TBE) belongs among the dangerous vector-borne diseases. The number of TBE incidences has been permanently increasing in various geographical regions, including the Czech Republic.
The presence of ticks and related diseases is driven by host-pathogen systems. The systems are rather complex and susceptible to environmental conditions represented in the first place by land cover/land use categories.
The presented study looks for a possible relation between the types of forest vegetation specified in the Landsat 5 satellite data and relative TBE morbidity. First, supervised classification of forest areas into five vegetation classes predefined by a botanist was tested.
Due to the spectral similarity of the classes, the resulting classification accuracy of Landsat scenes covering the entire area of the Czech Republic was quite low. Thus, an unsupervised approach was applied using nine spectral classes.
Relative TBE morbidity data collected over 10 years for 206 administrative units covering the entire country presented field data that were correlated with the spectral classes. The TBE risk index (IRE) of a given spectral class was introduced at each satellite scene.
To create a map of the TBE risk for the entire country, all IRE values were accumulated and divided into six risk categories. The disadvantages of the proposed method, especially regarding the accuracy of the final product with a nationwide cover age, are discussed.
In addition, the correlation between the relative TBE morbidity and other environmental parameters, such as annual precipitation, average temperature, and number of hunted game were calculated, but they did not reveal any significant relationship.