Change Detection Using Mappable Vegetation Related Indices A Case Study from Sinharaja Forest Reserve

Abstract

This study evaluates multi-year changes of vegetation in the Sinharaja forest reserve usingmappable vegetation related indices viz., Normalized Difference Vegetation Index (NDVI)and Burn Index (BI). Land-cover changes in the Sinharaja forest reserve region are detectedusing Landsat 7 ETM+ images for 1993, 2001, and 2005. Seven individual bands of eachimage were converted to new multiband files by layer stacking using ENVI® 4.5. Then themultiband files were re-projected to UTM Zone 44 North, WGS-84 Datum. Each data set wasexported to ENVI® EX software package to detect the changes between time steps based onNDVI and BI using an image difference tool. Land-cover data, which were obtained from theDIVA GIS web portal were compared with Landsat image data. Results of BI showed that theforest reserve fringe was vulnerable to forest fire. For example, 1993- 2001 period 160 haidentified as burn area, while it was 79 ha in 2001-2005 and 10 ha for the entire period of1993-2005. NDVI resulted in a 962 ha increase of vegetation prime at the western Sinharajain 2001-2005 periods. In addition, a 15 ha decrease in vegetation for 1993-2005 periods. Theresults are visualized using an embedded 3D render window of Google Earth and 2D view ofArcGIS explorer online. In conclusion, in-situ ground truthing data for the fire-influencedarea is important to implement sustainable forest resource management at the Sinharaja forestreserve.