Shaping smarter prediction: A literature review of GIS, remote sensing and machine learning applications in digital soil organic carbon mapping

Soil organic carbon (SOC) is an important indicator of soil health. Its regular monitoring helps achieve sustainable soil management, therefore, ensuring high agricultural productivity. The predicted SOC maps are becoming crucial and more important to monitor the climate change mitigation, sustainable environment. These particular needs have incited scientists and researchers to think about smart mapping approaches using new information technologies. This article aims to provide an overview of how GIS and remote sensing, modeled using machine learning algorithms, can contribute to SOC prediction for better understanding its spatial distribution across landscape. The review highlights the important role of Remote Sensing, Geographic Information System-Machine Learning (RS-GIS-ML) technologies. These approaches benefit from both field-collected data and features derived from spectral imagery acquired through various sensors (Multispectral, Hyperspectral, LiDAR and Radar). ML algorithms further enhance this process by modeling the complex relationships among multiple input variables, enabling the accurate SOC content in the form of spatial maps. The structure of this review is organized follow three principal sections: (1) use of remote sensing in SOC prediction, (2) machine learning methods for SOC prediction and (3) the integrated RS-GIS-ML approach for SOC mapping. The article concludes with a discussion on the future prospects of digital SOC mapping using geospatial technologies and ML algorithms.