38 / 2023-12-23 15:11:36

Land Use and Land Cover Change Detection Using Satellite Remote Sensing Techniques in Pao River Basin, Venezuela

land use land cover,satellite remote sensing techniques,tropical basin

In this investigation, land use and land cover (LULC) change detection were applied on a tropical basin, ten change detection methods have been evaluated; nine of these were classified like a methods based on pixels and the last corresponded to the classified object method. Eleven images were acquired from the Landsat satellite in period between 1986 and 2016. The satellite data acquisition was carried out from earthexplorer website. Eleven Landsat images were acquired; where the Pao river basin was contained. The scene was identified under the world reference system corresponding to the raw (005) and path (053).  The temporal series of images from three Landsat satellite can be grouped as follows: 1) L5TM (1986, 1990, 1991, 1998, 2001), 2) L7ETM (1999, 2000, 2002, 2003) and 3) L8OLI (2015 y 2016). The criteria for selecting of the temporal series of Landsat images were the same season of each year, and lowest coverage of clouds, aerosols and haze. Reference data were depicted by Google Earth images. The percentages of change area according to each change detection method of pre-classification were: a) Image difference: 7 to 10%, b) Image Ratioing: 0.5 to 3%, c) NDVI image difference: 1 to 4% and d) Principal Component Image Difference: 4 to 10%. Post-classification methods contributed with the pre-classification methods in a better approximation to the area difference proportion associated to each land use / land cover occurred in the study zone. Among the post-classification methods, it was found that the support vector machine provide results more approximates between these and their accuracy indexes were upper than those obtained through the maximum likelihood algorithm. Ten change detection methods were evaluated; nine of these were classified as methods based on pixels and the last corresponded to classified object method. Change detection method based on the pixel with the most capability for estimating LULC changes was the principal components using the component N° 1 compared with the rest of methods such as image difference, image ratioing, image regression and normalized difference vegetation index. This method included the most variance in the reflectance values in the visible and infrared spectral regions about LULC (agricultural, rangeland and urban likewise covers as water, vegetation and degrade soil). This method can be complemented with the image ratioing and change vector methods for achieving a better detection in those changes associated to a cover transformation from water to vegetation, and in the opposite sense.