CAPABILITY LANDS AT SATELLITE DATA FOR DETERMINATE CHANGE DETECTION IN FOREST COVER (CASE STUDY: KHORRAMABAD FORESTS, IRAN)
Assistant Prof., Dept. of Forestry, Faculty of Agriculture and Natural Resources, Sari University, Iran
|Corresponding Author: Hassan Akbari, E-mail: email@example.com|
|Received: 04 February 2014 Accepted: 28 February 2014|
|Related article at Pubmed, Scholar Google|
Large scale forest type mapping using current field methods is time consuming and cost-intensive. Satellite imagery capability in order to managing and mapping forest-covered area is a new useful tool. The purpose of this research was forest cover changes detection using satellite data from 1991 to 2003 and the assessment of capabilities different classification methods for Northern forests of Korramabad region located in Lorestan Province, Iran. In this research, the multi-spectral and panchromatic TM and ETM+ images has been used for forest area mapping. The images were geometrically corrected and orthorectified using GCPs and DEM. For comparing change detection in the forest, two methods including Normalized Difference Vegetation Index (NDVI), and different supervised classifications methods were used. The classification including forest/non-forest classes were accomplished by maximum likelihood, parallelepiped and minimum distance classification method. The accuracy of classification results were assessed with ground truth maps. The ground truth maps including forest/non-forest and reforested area maps had generated by interpretation of digital orthophotomosaic and field check. The results showed that forest and non-forest areas classification based on supervised classification method was more accurate than other methods. Using maximum likelihood algorithm, the best results of the classification was obtained from TM with 90.08% overall accuracy and 0.78 Kappa coefficient and ETM with 90.05% overall accuracy and 0.87 Kappa coefficient. The change detection results showed that the forest cover area has been decreased about 8.097% during 12 years. In spite of high overall accuracy and regards to the moderate Kappa coefficient, this research showed that investigation on forest cover change detection using satellite images doesn’t involve beneficial results.