Weather data-based prediction of Siberian moth Dendrolimus sibiricus Tschetv.: a case study
D. A. DEMIDKO, A. A. GOROSHKO, S. M. SULTSON, N. N. KULAKOVA, P. V. MIKHAYLOV
Reshetnev Siberian State University of Science and Technology, Krasnoyarsk, Russia
Keywords: southern taiga, Siberian moth, outbreaks, prediction, weather, machine learning
Abstract
The outbreak prediction is one of crucial components of forest pest management. Weather plays substanial role in the leaf-eating insects outbreaks esteblishment. The weather-based prediction models in this field are numerous and more or less precise. We attempted create such model for Siberian moth (Dendrolimus sibiricus Tschetv.) - one of most harmful defoliator in southern taiga of Siberia. For territory of interest (southern taiga and hemiboreal forests of Tomsk Oblast, Kemerovo Oblast and Krasnoyarsk Kray) the gradient boosting (XBGoost) model was created with accuracy 0,952. The temperatures of vegetation period 4th and 5th years before onset of outbreak are better predictors.
|
