MULTIDIMENSIONAL CLASSIFICATION OF ACTIVE-PASSIVE REMOTELY SENSED DATA FOR MONITORING OF HAZARD PHENOMENA OCCURRING ON DRAINED SOILS

摘要

Nowadays some of the vast-territories of Ukraine are liable to excessive wetting of swampy lands (which had been drained earlier on) as well as to spontaneous ignition of peaty soils. In this context it is quite a challenge to monitor these disastrous processes unless special-purpose facilities for remote sensing of the Earth are in active use. The vast areas can be efficiently monitored exclusively through the use of the state-of-the-art computer-based identification of hazard phenomenon manifestations. The monitoring operations are carried out using the aerial surveying data gleaned by active-passive tools for remote sensing of the Earth. The techniques for 3-D clusterization of data on the consecutive day-night thermal-radar airborne sensing of heavily wetted formerly drained soils and on subsurface self-heating of organic matter were proposed and tired out. The results of 3-D data clusterization suggest that the clusters are dependably separated and the clusters thus obtained are positively correlated with expert assessments of original data. It is shown that in order for the above techniques to be applied to other territories it is necessary that they should be optimized involving the use of remotely sensed and field data gathered from the analyzed art territories.