On correcting the time-lag bias in aerial-surveyed bark beetle infestation data

Abstract

Bark beetles are among the main biotic hazards affecting forests worldwide. Aerial surveys are an appropriate and commonly used tool for detecting and monitoring bark beetle infestations at the landscape scale. However, a considerable time-lag in the aerial-surveyed data limits its use, e.g. for analyses which include temporary factors.This temporal lag is caused by the gradual decay of the host tree following a bark beetle attack. A change in the spectral signature in aerial imagery indicates an attack with a delay of several weeks or even months. Thus, aerial-surveyed infestation data suffer a temporal shift, i.e. infestations detected for a specific time period cannot be directly assumed to have been initiated within that same period. The present study introduces an algorithm which assigns the most probable time of Ips typographus L. attack to infestations detected by aerial surveys. To calculate the infestations initiated within a certain year x the algorithm considers the potential attack period, the assumed time-lag, as well as the date of aerial surveys and the amount of detected infestation from the years x, x−1 and x+1. The deviation between infestations detected by aerial survey and those actually initiated (as calculated by the algorithm) quantifies the correction effect, i.e. the time-lag bias. The algorithm was then applied to a data set which covers a time series of 26 consecutive years, including both managed and non-managed infestations. To account for uncertainty in the input variables or to mimic varying environments the application considers three different scenarios. Over all three scenarios results indicate substantial deviations, which may be either positive or negative. Further, deviation is shown to be affected by the length of the time-lag and the attack period, as well as by variations in survey dates and infestation frequency of consecutive years. Finally, the time-corrected infestation data provides a threefold benefit: (i) The temporal pattern of oscillating population dynamics is truly reflected. (ii) Comprehensive correlation analyses which include dynamic environmental covariates, e.g. temperature, precipitation, or antagonists, are facilitated. (iii) Aerial-surveyed and ground-surveyed monitoring data can be easily related whether within a region or between regions. In conclusion, this study increases the value of aerial-surveyed infestation data for bark beetle monitoring and research.