چكيده لاتين
Vegetation indices are criteria used to measure and assess the condition of vegetation. These indices have also been developed to differentiate and extract vegetation cover on the Earthʹs surface using satellite images. Accordingly, many indices have been proposed, one of which is the Enhanced Vegetation Index (EVI), suggested as an improved version of the Normalized Difference Vegetation Index (NDVI) to reduce the effects caused by the atmosphere and background soil. The EVI2 is also used for sensors that do not have a blue band. One of the challenges ahead is the sensitivity analysis of the EVI and EVI2 vegetation indices to effective parameters, including spatial resolution, vegetation cover, topography type, and seasonal changes. As known from previous studies, fixed coefficients calculated from MODIS sensor images are used to compute the EVI and EVI2 vegetation indices. The main objective of this thesis is to examine the impact of effective parameters on the two vegetation indices, EVI and EVI2, to determine whether fixed coefficients can be used under all conditions or if it is better to calculate optimal coefficients for each image based on its conditions. In this research, Landsat 8, Sentinel 2, and MODIS images were used due to their availability and accessibility. Considering that one of the environmental parameters under investigation is seasonal change, similar images were obtained for each season, and after performing corrections suitable for each image, three ranges of the corrected image for vegetation cover were extracted using the NDVI, including less than 10% cover, between 10% and 20%, and more than 20%, along with three ranges for topography classes 1, 2, and 3. Therefore, in the end, for each of these, 24 images and a total of 72 images with different spatial resolutions, vegetation cover, and topography were tested across four seasons. Then, for each of the images, a reference map including two classes of vegetation cover and non-vegetation cover was prepared using the SVM classifier method in ENVI 5.3 software. To calculate the optimal coefficients, the PSO algorithm was used due to its simplicity, high convergence speed, and fewer computations.
Ultimately, in the 72 images examined, the standard coefficients in the EVI vegetation index showed satisfactory performance for only 33% images, while in the other 67% images, the PSO algorithm was able to calculate coefficients that yielded better results than the standard coefficients. In the EVI2 index, 35% images using standard coefficients and 65% images using optimal coefficients showed satisfactory performance, indicating that the number of images in both indices suggests that the use of standard coefficients is not suitable under all conditions, and it is better to calculate coefficients according to the conditions of each image and then use them in the relevant index.
