Many methods for trajectory simulation, such as Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT), have been developed over the past several decades and contributed greatly to our knowledge in insect migratory movement. To improve the accuracy of trajectory simulation, we developed a new numerical trajectory model, in which the self-powered flight behaviors of insects are considered and trajectory calculation is driven by high spatio-temporal resolution weather conditions simulated by the Weather Research and Forecasting (WRF) model. However, a rigorous evaluation of the accuracy of different trajectory models on simulated long-distance migration is lacking. Hence, in this study our trajectory model was evaluated by a migration event of the corn earworm moth, Helicoverpa zea, in Texas, USA on 20–22 March 1995. The results indicate that the simulated migration trajectories are in good agreement with occurrences of all pollen-marked male H. zea immigrants monitored in pheromone traps. Statistical comparisons in the present study suggest that our model performed better than the popularly-used HYSPLIT model in simulating migration trajectories of H. zea. This study also shows the importance of high-resolution atmospheric data and a full understanding of migration behaviors to the computational design of models that simulate migration trajectories of highly-flying insects.
Deep Q-Learning for Two-Hop Communications of Drone Base Stations
