A comprehensive analysis of the use of modelling and remote sensing techniques for monitoring and managing rangelands

Sustainable rangeland management supports livestock production, food security, and key ecological services such as carbon sequestration and water regulation. However, rangelands face increasing pressure from climate change, land degradation, and agricultural expansion, requiring effective management strategies. This review follows the PRISMA guidelines and systematically examines 102 peer-reviewed publications selected from 511 initially identified studies across multiple databases, including Scopus, Google Scholar, ScienceDirect, AJOL and Web of Science. This review explores the latest tools enabling accurate monitoring and prediction of rangeland dynamics. The results show that key technologies include machine learning algorithms, unmanned aerial vehicles (UAVs), and multispectral sensors, all of which have revolutionized biomass estimation. Satellite remote sensing, particularly Sentinel-2 and Landsat 8/9, represents a transformative advancement by delivering consistent, scalable, and repeatable observations from regional to global scales. Methods such as Deep Neural Networks (DNN), Random Forest (RF), and Object-Based Image Analysis (OBIA) have outperformed conventional algorithms, achieving performance metrics such as R 2 > 0 . 85 . Generalized Linear Models (GLM) have also been widely applied, particularly for environmental impact assessment. The development of multispectral sensors, especially bands such as NIR and red-edge, has improved vegetation index calculations, while LiDAR technology has enhanced biomass prediction by incorporating terrain structure and canopy height data. Despite these advances, challenges remain, including issues related to data quality, sensor integration, and harmonizing datasets for predictive modelling. This review highlights both the strengths and limitations of current approaches and emphasizes the need for further integration of advanced technologies such as hyperspectral sensors. • Rangeland productivity is better predicted using vegetation indices and models. • Machine learning improves monitoring of drylands using satellite observations. • Integration of field and remote sensing data enhances forage yield estimation. • Random Forest models perform well for biomass mapping in rangeland systems. • Research gaps identified for future remote sensing and modelling applications.