Detection of Vegetation Changes in Agricultural Lands of Sistan Plain, using Remote Sensing Technique

Abstract

Information about the state of vegetation is very important for environmental planning, land preparation and achieving sustainable development. In this study normalized differential vegetation index (NDVI) values were calculated based on Landsat 8 satellite images in order to show temporal and spatial changes in the vegetation cover of agricultural lands in Sistan plain over ten years (2011 to 2020) using the Google Earth Engine platform. Additionally, the NDVI index were classified using decision tree algorithm in order to analyze vegetation changes using thematic change workflow method. By comparing classified images with reference samples which collected from ground sampling, validation was carried out. Then, in order to assess accuracy of vegetation maps, the error matrix was prepared, the overall accuracy and kappa indices were determined. The values of overall accuracy and kappa indices indicated optimal accuracy and it can be stated that there is moderate agreement between ground samples and the classified images (i.e., kappa index is 0.48 to 0.7). The central areas of Sistan plain have a decline in vegetation, whereas areas in northern and eastern have an increase. The cover vegetation on lands of Sistan plain decreased in 19260.4 ha while increased over 25633.2 ha throughout ten years. Examination of NDVI index shows instability of production in this area due to aforementioned factors.

Keywords

• vegetation indices
• remote sensing
• environmental monitoring
• google earth engine
• landsat image