An efficient method to find area clusters with constraints using grid index structure

This work presents an efficient method to find area clusters satisfying certain constraints. The method uses a grid index structure to find them by examining five cell, 8-connected neighborhood. The hierarchical clustering technique is applied for preprocessing the grid with similar value distribution into the same clusters. This preprocessing helps reduce the computational time and eliminate the possible outlier grid cells. The constraints are average value range, minimum area size and minimum missing data ratio. The method is implemented for a remote sensing data, MOD08_M3, which is MODIS (MODerate resolution Imaging Spectroradiometer) level 3 monthly atmospheric product.