Employing the agricultural classification and estimation service (ACES) for mapping smallholder rice farms in Bhutan

Creating annual crop type maps for enabling improved food security decision making has remained a challenge in Bhutan. This is part due to the level of effort required data collection, technical model development, and reliability an on-the-ground application. Through focusing on advancing Science, Technology, Engineering, Mathematics (STEM) Bhutan, co-develop geospatial application known as Agricultural Classification Estimation Service (ACES) was created. paper focuses co-development Earth observation informed climate smart framework which incorporates both modeling training sample collection. The ACES web subsequent software package enables stakeholders more readily use into their process. Additionally, this offers transparent replicable approach addressing combating remote sensing limitations topography cloud cover, common problem Lastly, resulted Random Forest “LTE 555” model, from set 3,600 possible models, with overall test Accuracy 85% F-1 Score .88 2020. independently validated resulting independent accuracy 83% .45 insight perturbation via hyperparameter tuning input features key future practitioners.