Abstract:
In response to the persistent challenges posed by fall armyworm (FAW) outbreaks in Rwanda’s maize production since 2017, this research introduces an innovative strategy integrating fuzzy logic and neural network methodologies, originally developed for hydrometeor identification. The focus is on distinguishing flying adult FAW moths using four polarimetric radar parameters: horizontal reflectivity (DBZHC), correlation coefficient (RHOHV), differential reflectivity (ZDR), and specific differential phase (KDP). Demonstrating a remarkable accuracy with a fraction of echoes correctly identified (FEI) of 98.42% for FAW and 87.02% for other weather phenomena, validated by a Heidke skill score (HSS) of 0.9801, the system proves adept at discerning between weather and nonweather events. A significant strength of the developed method lies in its ability to detect FAW adult moths approximately four weeks earlier than ground-based observations identifying infestation outbreaks, This was evident in the context of FAW infestation in maize fields within the surveyed districts of Nyanza, Huye, and Gisagara in the Southern Province of Rwanda. This positions the weather radar method as a promising early warning system for FAW outbreaks, especially beneficial in less-monitored regions like East Africa. The study underscores the potential application of polarimetric C-band Doppler weather radar, providing valuable insights into the intricate dynamics of agricultural insect pest outbreaks. The method offers practical solutions for timely interventions and enhanced crop management strategies, contributing to more effective pest control and promoting sustainable agriculture practices.