Land Cover Classification and Forest Normalized Difference Vegetation Index (NDVI) Analysis of Manaslu Conservation Area

Abstract

Remote sensing is nowadays widely used in study and management of environment both in spatial and temporal scales. Land cover classification is one of the earliest applications of remote sensing. Remote sensing can also be used in understanding different ecological phenomenon. This work encompasses land cover classification and productivity (NDVI) analysis of five high mountain forests of Manaslu Conservation Area (MCA) of central Nepal using remote sensing and GIS. Normalized Difference Vegetation Index (NDVI) calculated from remote sensing image is an indicator of vegetation vigor, productivity and health. It is the simplest index to understand the vegetation performance in different scales. Landsat ETM+ image is used to classify land cover. Unsupervised and supervised methodology was used for classification in ERDAS imagine software. Accuracy of classified map was assessed by confusion matrix. NDVI analyses of five different forest patches (Betula-Abies, Larix, Pinus wallichiana, Quercus and Picea-Tsuga) were done using MODIS terra data products. NDVI of each forest patches was acquired from 16 days composite 250 m MODIS data of year 2000 January to 2008 December. Relation between NDVI and total monthly precipitation and average temperature of nearest weather station (Gorkha) was also tested. Land cover map is acquired with 60.14 percent of overall accuracy. Boulder & Grass occupies highest area in MCA followed by human influenced land cover Agriculture & Settlement. Among five forest examined Picea-Tsuga forest is found with highest NDVI followed by Quercus forest. Betula Abies forest of highest altitude is found with lowest average NDVI value. Larix and Pinus wallichiana forests lie in between them. Except maximum value of Larix no forest showed any trend of increasing or decreasing NDVI from 2000 to 2008. One month lag of average monthly temperature and two month cumulative rainfall has been found as best predictor of NDVI for most of the forest types. Temperature is linearly related to NDVI and is seen as limiting factor for productivity of high mountain forest. Precipitation, however is unimodally related to NDVI exhibiting highest NDVI in moderate rainfall.