Forest damage assessment in the Black Triangle area using landsat TM, MSS and forest inventory data

Abstract: The main objective of this work was the evaluation of TM (Thematic Mapper) and MSS (Multi Spectral Scanner) data of Landsat 1 and 5 for a large scale forest damage assessment in the black triangle region. The area covers the Krusne Hory and neighboring mountain areas (border area of Germany, Poland and the Czech Republic. A temporal comparison of land use of 1975 and 1989 of a sub area and a classification of three forest damage categories of coniferous forest was accomplished. A geometrical corrected database was produced and TM data was radiometric corrected to at satellite spectral reflectance. To achieve more comparable TM data sets(3), a relative atmospheric correction of one TM data set was performed using forest and lake test sites in overlapping areas of the data and a linear regression algorithm. A land use classification of TM and MSS data was performed using a supervised maximum likelihood classifier. Logit regressions for an estimation of the degree of needle loss were used to discriminate among 3 damage categories. The results indicated that a relationship between spectral properties and forest damage exists. The land use classification showed an increase of clear cuts from 1975 to 1989 and a decrease of coniferous forest. Classification results verified that atmospheric conditions in the study area had an effect on the accuracy that was achieved. It was concluded that large areas (more than one TM data set)(4) are not accurate to classify without additional ground truth data for every data set. (3) TM data sets used for this work contained a sub area of channel 1 to 5 and channel 7. (4) For this work subareas of 3 TM data sets were used.