Modeling weather-driven long-distance dispersal of spruce budworm moths (Choristoneura fumiferana). Part 2: Flight model calibration using radar data

Abstract

In Part 1 of this series (Garcia et al., 2022), we introduced a novel individual-based model for the simulation of dispersal flight of adult spruce budworm (SBW: Choristoneura fumiferana) and demonstrated the results of that model under real weather conditions for two nights in July 2013 on which SBW mass dispersal events were observed by weather radar in southern Quebec, Canada. Here, following the selection of one uncertain parameter value using empirical measurements, we used those radar observations for the quantitative calibration of two uncertain flight model variables in our individual-based SBW–pyATM model, one that describes the conversion of moth wingbeat to flight speed, and a second that allows the moth to conserve energy during flight. For these experiments, we adapted a grid-based metric from meteorology that has previously been used to calibrate and validate precipitation forecasts by comparison with radar data. Through thousands of flight simulations for the night of 15–16 July 2013, examining each of these parameters separately and in conjunction, we arrived at optimal values that produce a spatiotemporal distribution of SBW moth dispersal that most closely matches the radar observations for that night. We then applied those calibrated parameter values to simulations of SBW dispersal on the night of 14–15 July 2013 and found a lesser but still reasonable resemblance to weather radar observations on that night as well. These two parameters have significant effects on the speed, altitude, and distance of dispersal, and are thus critical to the goal of estimating when and where SBW males and females land, with subsequent effects on reproductive behavior and the spatial redistribution of SBW populations over a dispersal season.