Satellite Imagery Analysis for Climate Monitoring and Space Applications Using Multi-Source Data Fusion and Deep Learning

Abstract

Big climate analysis, space evaluation and space analytics involves an inseparable element of satellite data. The complex of the Artificial Intelligence (AI), Machine Learning (ML), and Deep Learning (DL) products has contributed significantly to the improvement of the applications of remote sensing in the context of automation and precision. Convolutional Neural Networks (CNNs), spectral-spatial as well as hybrid deep learning models have been found to be better when it comes to classification and mapping land cover activities in the environment [1], [12]. Transformer-based models and CNN-Transformers hybrids are also used as latest developments and are more efficient at deriving global contextual features in satellite images [21], [22]. The detection of floods, monitoring of glaciers, mapping of wild fires, analyzing the urban heat, and numerous others are only a small number of real-world applications of AI-powered satellite systems [6], [9], [10]. Despite these advances, the problems of computational complexity, multi-sensor data heterogeneity, small scale real-time processing and cross-regional generalization [5], [18], [20] remain present. Research gaps, methodological tendencies, identification of 22 research studies, systematic review, forming the conceptual basis of an integrated climate monitoring system dubbed GeoSpectra is presented.