Coastal environments are always under the pressure of natural processes such as erosion, sedimentation, natural disasters as well as human projects. These threats have made coastal areas a priority for coastline monitoring and sustainable coastal management programs. In this paper, algorithms for separating water and land boundaries as well as new sub-pixel methods are presented with the aim of dividing large pixels (with low resolution and spatial accuracy) into smaller pixels and creating a classified map with better spatial resolution. Different water identification indices and machine learning algorithms were investigated, and two models of Spatial Attraction Models were implemented. Results showed that the Sub-pixel / Sub-pixel Spatial Attraction Model had more capacity in providing higher resolution and precision, while provided 10% reduction in error when compared with observations. To skill assess these two methods, the difference in areas created by each method compared to the reference shoreline (high resolution aerial image) was computed. Also, in order to accurately evaluate and show the high accuracy of sub-pixel algorithms, the results of these algorithms should be examined by conventional classification methods. The creation of such models is proposed to support integrated coastal management in the Persian Gulf region for future studies.