A STUDY ON VARIOUS CHANGE DETECTION TECHNIQUES USING DIGITAL REMOTE SENSING DATA

Abstract

Land is an important natural resource. Its proper and effective utilization is extremely important to a country like India. The changes in various earth features and natural resources may have an impact on social and economic development of the region. Therefore, it is very important to monitor these- changes over a specified interval of time. Various change detection techniques have already been developed for 'this purpose. Among those, the techniques which use digital data are found to have more advantages than the techniques using hard copy data. .Roorkee and its surrounding areas lying in the U.P. State of India has been selected as the study area. In the present study, various change detection techniques viz., image differencing image regression, image ratio and NDVI differencing are considered using the digital satellite data of LANDSAT -MSS, 1974 and IRS-LISS II, 1989. The study has been carried out in two stages. In.the first stage, digital change 1 detection techniques using different bands and threshold values are employed to study the behaviour of various change-detectian techniques for monitoring the changes in the study area. Only absolute changes have been considered here. In the second stage, satellite data of the years 1974 and 1989 are classified independently using Maximuil Likelihood Supervised Classifier. Tho classified images are then compared indirectly with the help of change maps obtained from the four techniques mentioned above. The assessed changes are then presented into the form of a change matrix, which shows the distribution of various land use changes between two time periods. Image regression and image differencing techniques with lower threshold values 1.25a) are found to be more useful for monitoring the changes using band 1-or band 2 data. While using band 3.oT band 4, image ratio technique with higher threshold values (>1.50a) is able to identify changes accurately when compared with the differencing techniques. However, .1,11)VI differencing using band 2/3 or 2/4 is able to identify more changes when compared with image ratio technique using band 3 or band 4 data. Further studies are also suggested to. .be carried out by using various techniques in entirely different terrain:and environmental conditions, so that the suitahility, of each ,change detection technique caa'be ortahl ished.