European Journal of Sustainable Agroecosystems

AGB (aboveground biomass) is a popular indicator of maize (Zea mays L.) growth status, nutritional sufficiency, and yield potential. Traditional methods for obtaining AGB are time-consuming, difficult, and not well-suited for precision agricultural use. UAV remote sensing has quickly emerged as a promising alternative, resulting in research studies evaluating LiDAR, hyperspectral imaging, and GIS spatial-analysis separately and combined for non-destructive AGB estimations. This review summarizes peer-reviewed studies about combining UAV-mounted LiDAR, hyperspectral sensing, or GIS spatial analysis for AGB measurements. LiDAR canopy height, canopy volume, and point clouds are employed for three-dimensional biomass accumulation data but do not provide information about pigments, water, or nitrogen that can be achieved through hyperspectral data acquisition. GIS methods allow for geostatistics-based interpolation and creating continuous GIS biomass layer. The studies reviewed show that the structural models explain 60-84% of the variance in AGB. The spectral models account for 50-70% of AGB variability. Studies using combination of LiDAR, hyperspectral and GIS data yield increasingly accurate data, with the coefficient of determination of R2>0.85 and lower RMSE than the data obtained using single methods. Machine learning methods such as random forest, support vector regression, and partial least square regression The study shows that the combination of information from various sources enables successful performance of random forest across cropping stages. The article highlights some ongoing issues - high cost of sensors and their payload, effects of canopy saturation in advanced stages of growth, problems with the transfer of models, and processing of complex point-cloud data and hyperspectral data. The article summarizes that, although UAV-GIS integration is a reliable scientific way to monitor biomass, its practical usage for efficient nutrient management, irrigation scheduling, and crop yield forecasting will depend on the advent of standardized data acquisition processes and cross-environment test trials.