Land Cover Mapping Supervised Classification

14 According to Yazidhi 2003, the use of digital elevation models and GIS offers possibilities to estimate more relevant topographical parameters that are useful in soil erosion modeling.

2.4.1. Land Cover Mapping

Land cover mapping is one of the most important and typical applications of remote sensing data. Land cover corresponding to the physical condition of ground surface, for example, forest, glass land etc., while land use reflects human activities such as the use of the land, for example, industrial zone, residential zone, agricultural fields etc. To prepare, the land cover mapping from digital images “land cover classification” should be done. There are two kinds of classification, i.e. supervised and unsupervised classification.

2.4.2. Supervised Classification

Supervised classification is the method used to transform multi spectral image data into thematic information classes. This procedure typically assumes that imagery of a specific geographic is gathered in multiple regions of the electromagnetic spectrum. In supervised classification, the identifying and location of feature classes or cover types urban, forest, water, etc are known beforehand through fieldwork, analysis of aerial photographs, or other means. Typically, identify specific areas on the multispectral imagery that represent the desire known feature types, and use the spectral characteristics of theses known areas to train the classification program to assign each pixel in the image to one of these classes. Multivariate statistical parameters such as means, standard deviation, and correlation matrices 15 are calculated for each training region, and each pixel is evaluated and assigned to the class to which it has the most likelihood of being a member according to rules of the classification method chosen. One of the sample classification strategies that may be used is Maximum Livelihood Classifier. The maximum livelihood was adopted by using the training samples of the landsat image and ground truth. Actually, this is one of the most popular methods of classification in remote sensing, in which a pixel with the maximum likelihood is classified into corresponding class.

2.5. Runoff Erosion Potential with GIS Based Spatial Analysis Model Spatial Analysis