Abstract:
Desert locust (Schistocerca gregaria) plagues threaten agricultural production, food security and the environment across Africa, the Middle East, and Southwest Asia. Control methods targeting adult desert locusts present significant challenges and financial costs. Recognizing this, we developed a ground-breaking fuzzy set Mamdani type inference model that provides an innovative solution for early warning alerts. The model aids in predicting the juvenile stages of locust development, thereby preventing wide-scale locust swarming and mitigating its
extensive damages and socioeconomic costs. The novelty of our approach lies in our unique application of environmental variables relevant for locust breeding to estimate the timing and location of desert locust hatching. Additionally, we improved the algorithmic handling of these variables, with localized desert locust bands data used as a proxy for hatching timing with a temporal offset of 35 days. The model’s boundary conditions were determined using a training area in Sudan, where comprehensive ground data was available. This rule set was subsequently applied to Turkana County in Kenya, a data-scarce region, demonstrating the model’s
applicability and success in different contexts. The model’s accuracy, assessed by data from the Sudan training site, demonstrated a remarkable score of 82% for true predictions. Furthermore, the model correctly identified the months of highest hatching probabilities in Turkana during 2020, demonstrating its real-world effectiveness and practical value. A correlation analysis affirmed that hatching was associated with increases in chlorophyll levels and precipitation accumulations. Our study marks a significant advancement in predicting the timing of
hatching using fuzzy logic in data-scarce environments. By operationalizing more targeted early responses to desert locust infestations, our model facilitates more effective locust control. This study stands as an important contribution to locust management strategies, with substantial implications for agricultural production and food security in affected regions.