Agriculture is a critical source of employment in rural Colombia and is one of the sectors most affected by climate and climate change and where solutions to key challenges affecting the productivity and sustainability of forages and the livestock sector are required. Increasing yields of forage crops can help improve availability and affordability of livestock products while also easing pressure on land resources through enhanced resource utilisation. This study aims to develop remote sensing-based approaches for forage monitoring and biomass prediction at local and regional levels in Colombia. Local access to such information can help improve decision making and increase productivity and competitiveness while minimising impacts on the environment. Ten locations were sampled between 2018 and 2021 across climatically distinct areas in Colombia, comprising five farms in Patía in Cauca department, four farms in Antioquia department, and one research farm at Palmira in Valle de Cauca department. Ash content (Ash), crude protein (CP %), dry matter content (DM g per square meter) and in-vitro digestibility (IVD %) were measured from different Kikuyu and Brachiaria grasses during the field sampling campaigns. Multispectral bands from coincident Planetscope acquisitions along with various derived vegetation indices (VIs) were used as predictors in the model development. To determine the optimum models, the improvement capabilities of using an averaging kernel, feature selection approaches, various regression algorithms and metalearners (simple ensembling and stacks) were explored. Several of the applied algorithms have built-in best feature selection functions so to test model improvement capabilities of an independent feature selection approach for algorithms that have one built-in and for those that do not, all models were run a) with no feature pre-selection, b) with Recursive Feature Elimination (RFE, package: caret) and c) with Boruta (package: Boruta) feature selection. A range of algorithms (n=26) belonging to classes of decision trees, Support Vector Machines, Neural Networks, distance-based methods, and linear approaches were tested. All algorithms including metalearners were tested with each of the three feature selection approaches while employing 10-fold cross-validation with 3 repeats. In the performance evaluation based on unseen test data, CP and DM was predicted relatively well for all three sites (R2 0.52 – 0.75, RMSE 1.7 – 2.2 % and R2 0.47 – 0.65, RMSE 260 – 112 g/m2 respectively). As part of the study, the investigation was carried out in cooperation with smallholder farmers to determine their attitudes and potential constraints to mainstreaming such technologies and their outcomes on the ground. Through improving communication between earth observation and agricultural communities and the successful integration of satellite-based technologies, future strategies can be implemented for increasing production and improving forage management while maintaining ecosystem attributes and services across tropical grasslands.