Advances in Remote Sensing and Machine Learning for Land Use/Land Cover Change Detection and Climate Model Bias Correction: Methods and Applications

Abstract

ABSTRACT

Anthropogenic and climatic pressures now threaten ecosystems globally, most alarmingly in data-scarce regions like the Ethiopian Rift Valley where sustainable resource management is urgently needed. In this meta-analysis, we synthesize 44 peer-reviewed studies (2004-2025) that were identified and normalized using the PRISMA review process, exploring the integration of remote sensing (RS) and machine learning (ML) for land use/land cover change detection and climate model bias correction. This review provides a critical synthesis of advancements in these fields to address two interconnected challenges in environmental forecasting: change detection of land use/land cover (LULC), and bias correction of climate models. We systematically assess the progression of methodologies, ranging from conventional pixel-based change detection and statistical bias correction to contemporary deep learning and hybrid methodologies. The review affirms that although multi-sensor RS data and advanced machine learning (ML) algorithms function as powerful standalone tools, their transformative potential is fully realized through integration. Our analysis identifies a key integration gap: despite the standalone power of sophisticated algorithms, their full potential remains unexploited due to a lack of integrated frameworks. To fill this gap, we propose a novel, integrated pathway for developing end-to-end systems. We conclude that robust environmental prediction depends on multi-sensor data fusion and explainable AI (XAI) for developing transparent, trustworthy, and actionable science-based policy and management tools.

Keywords: Climate Model Bias Correction, Data Integration, Land Cover Change Detection, Machine Learning, Remote Sensing, Explainable AI