2. Location of the Study Area The Rangpur Sadar Upazila was selected as a study area, which is under Rangpur District; located in the northwestern part of Bangladesh. The entire District is about 2370.45 km2 in which Rangpur Sadar Upazila consists of 330.33 km2 area. Rangpur Sadar Upazila is situated in between 25°36´ N to 25°50´ N latitudes and 89°5´ E to 89°20´ E longitudes. This Upazila is bounded by Gangachara Upazila at north, Mithapukur Upazila at the south, Pirgachha Upazila at east and Taraganj Upazila at the west side. The population of Rangpur District is approximately 28,81,086; in which Rangpur Sadar Upazila contains about 25,1699 (BBS, 2011). The Teesta and the Ghaghat rivers are the main hydrological systems in the study area. Geologically, the area lies under north-northwestern part of Bengal Basin and falls on old Himalayan piedmont plain (Khan, 1991). The District intensively represents recent floodplain deposits consisting soil types of silt, clay, fine to medium-grained sands (Islam et al., 2014). The average maximum temperature of summer period is 32°C and average minimum temperature of winter period is 11°C. The annual average rainfall is 2931mm (BWDB, 2010).
3. Data Collection: Three multi-date geo-coded Landsat imageries of 1972, 1989 and 2014 were used to measure and mapping urban vegetation of Rangpur Sadar Upazila. Mainly the Blue, Green, Red and Near-Infrared bands from the each set ofimagery were selected to generate the Normalized Vegetation Index (NDVI) for the three dates. These geo-coded cloud free imageries which rectified to the UTM project system were collected from the United States Geological Survey (USGS) website as free of cost. To mask out the study area from the whole scene of Landsat, a vector polygon shape file was used. This shape was managed from the Bangladesh Local Government and Engineering Department (LGED). In addition to this data, temporal population of each union data was collected from the national census of the city in order to establish a regression model between the population data and declined vegetation data.
The major steps involved during the methodology stage were image pre-processing, generation of NDVI, perform change detection analysis, area measurement, statistical analysis as well as accuracy assessment.
4.1. Pre-processing of Landsat Data The image pre-processing comprises certain necessary preparatory steps such as radiometric correction, geometric correction and image enhancement in order to improve the quality of original images, which then results in the assignment of each pixel of the scene to one of the vegetation groups defined in a vegetation classification system (Das et al., 2013). Most of the pre-processing tasks: image loading, added in layers, generation of RGB, image enhancement, radiometric correction, NDVI generation, area measurement and accuracy assessment were completed using ENVI v 4.7 and ArcGIS 10 software.
4.2. Normalized Difference Vegetation Index (NDVI) NDVI is a slope based vegetation index widely used for extracting vegetation information from the earth using remotely sensed imageries. It is the most commonly used Vegetation Index (VI) as it retains the ability to minimize topographic effects while producing a linear measurement scale (Silleos et al., 2016). This algorithm mainly uses visible RED and Near-Infrared to indicate abundance of vigor green vegetation and biomass. On the other hand, it explores the existence of soil, rock, water and ice. The classic calculation of an NDVI is shown in equation (1). From this calculation, NDVI values close to 1 represent green vegetation while near to zero or minus show non-vegetation feature.
Many remote sensing researchers used NDVI for identifying and quantifying different vegetation features. Kumagai et al. (2008) used Landsat ETM+ data to generate NDVI to analyze seasonal fluctuation as well as spatial distribution of vegetation in Kansai District, the Western part of Japan. He integrated autocorrelation with NDVI to extract vegetation changes in the city in terms of seasonal variation. A significant variation in vegetation areas was found in his paper. Nichole et al. (2005) identified vegetation cover and vegetation density using NDVI in Kowloon city, Hong Kong. While vegetation density and vegetation coverage are important parameters for a city in terms of urban environmental quality, the city green vegetation has been reduced at a remarkable rate in their results measured.
Many remote sensing researchers used NDVI for identifying and quantifying different vegetation features. Kumagai et al. (2008) used Landsat ETM+ data to generate NDVI to analyze seasonal fluctuation as well as spatial distribution of vegetation in Kansai District, the Western part of Japan. He integrated autocorrelation with NDVI to extract vegetation changes in the city in terms of seasonal variation. A significant variation in vegetation areas was found in his paper. Nichole et al. (2005) identified vegetation cover and vegetation density using NDVI in Kowloon city, Hong Kong. While vegetation density and vegetation coverage are important parameters for a city in terms of urban environmental quality, the city green vegetation has been reduced at a remarkable rate in their results measured.
4.3. Change Detection Change detection is the process of identifying differences in the state of an object or phenomenon by observing it at different times (Singh, 1989; Nori et al., 2008). In this paper post classification, comparison method of change detection used to calculate spatial changes of the urban vegetation. After each NDVI classification, subtractions from 1972-1989, 1989-2014 and 1972-2014 were performed to calculate changes of vegetation area.
4.3. Change Detection Change detection is the process of identifying differences in the state of an object or phenomenon by observing it at different times (Singh, 1989; Nori et al., 2008). In this paper post classification, comparison method of change detection used to calculate spatial changes of the urban vegetation. After each NDVI classification, subtractions from 1972-1989, 1989-2014 and 1972-2014 were performed to calculate changes of vegetation area.