Variable-rate application of fertilisers is one of the agricultural techniques for site-specific management that emphasises the application of the optimal amount of fertiliser for each crop. This technique involves applying fertiliser only on areas that require it instead of the standard practice of uniform application. Traditionally, farmers would estimate the amount of fertiliser needed for different farm areas. Geospatial computing is a relatively new and advanced technology that allows farmers to use proximal, aerial, and satellite sensors to determine the state of a crop and the amount of fertiliser required for different areas. However, it has been challenging to utilise this technology on Tanzanian farms because of the cost of the equipment to do such variable fertiliser applications.
This research will explore smartphone usage and unmanned aerial vehicle (UAV) imagery to aid variable-rate fertiliser application. The study proposes building a model that will use vegetation indices and machine learning to predict the site-specific requirements for fertiliser and derive a prescription map that can be sent to farmers’ smartphone apps to assist them in implementing the variable-rate application of fertilisers. Controlled fertiliser application will save costs, boost production, and preserve the environment.