Improving the Temporal and Spatial Accuracy of the Normalized Difference Vegetation Index (NDVI) Map using Satellite Image Fusion Algorithms

In vast areas, accessing satellite images with appropriate spatial resolution, such as Landsat images, is often challenging.dditionally, the temporal resolution of the Landsat satellite does not allow for the examination of short-term changes in phenomena such as vegetation.The aim of this research is to utilize temporal and spatial fusion techniques of Landsat-8 and MODIS satellite images to prepare a Normalized Difference Vegetation Index (NDVI) map.For this purpose, six image fusion algorithms-NNDiffuse (Nearest Neighbor Diffusion), PC (Principal Component), Brovey, CN (Color Normalized), Gram-Schmidt, and SFIM-were applied in an experimental area in Khuzestan province.After evaluating the results of these algorithms and selecting the most appropriate algorithm based on statistical indicators (spectral criteria such as the correlation coefficient and spatial criteria such as the Laplacian filter), the spectral and spatial information from the red and near-infrared bands of eight mosaic Landsat-8 images (30 m resolution) were combined with the red and near-infrared bands of one MODIS image (250 m resolution).To investigate vegetation cover, the NDVI was calculated using the fused satellite image for Khuzestan province.The results showed that the NNDiffuse fusion algorithm demonstrated very high accuracy among the tested algorithms in terms of spatial evaluation and spectral quality criteria.Consequently, this algorithm was selected to combine the red and near-infrared bands of Landsat-8 and MODIS images.Compared to the original Landsat-8 image, the NDVI map prepared using this algorithm had the lowest statistical errors, with an RMSE (Root Mean Square Error) of 0.1234 and an MAE (Mean Absolute Error) of 0.081.