Validating algorithms designed for fertilization control in rice farming system

Rice yields in Rwanda seldom reach values expected for irrigated production, in part due to poor fertilization management. To adapt well-known best practices to local conditions, two fuzzy fertilization algorithms were designed for conditions of rice production in Rwanda. Fuzzy algorithms have a linguistic rule base of 183 IF THEN statements linking measurable field conditions to crop yield. Interviews with government agricultural experts and published knowledge of site conditions were used to create the rule base which incorporates the known nutrient requirements of the different growth stages of rice. Fuzzy algorithms are intended for use in an Internet of Things (IoT) system. To validate the algorithms, historical weather and field data are used to drive yield simulations for different plots during the first growing season of 2020 at sites in northeast Rwanda. Predicted yields are compared to measured yields for scenarios with different irrigation levels and fertilization amounts and with and without full Internet connectivity. Insights are provided on the potential of these fuzzy methods to improve rice nutrient management for increased yields in low-income countries are discussed.