Land Use and Land Cover (LULC) classification is considered to be the most significant anthropogenic strategy for the integrity of the ecosystem. In general, the LULC provides an effective support for land conservation and management. However, the traditional method requires more cost, especially for the data analysis and the classification. Satellite imagery data are used for determining the LULC map, which is useful for land management. The downstream analysis robustness has been enhanced by quantifying the pixel-level uncertainty. In general, most of the task related to this type of classification tends to lack uncertainty quantification and take more computational resources for large-scale analysis. Therefore, it is significant to develop a novel LULC classification framework by tackling multiple complications involved in the classical techniques. Here, essential satellite images are sourced from available dataset. Further, the taken satellite images are provided to the LULC classification phase. In this phase, a Multiscale Efficient Network with Sparse Attention (MENet-SA) is employed to execute the LULC classification tasks. Finally, LULC classification outcomes are attained from MENet-SA. Then, several validations are executed in the developed LULC classification model over classical techniques to verify its efficiency.