Monitoring the spatio-temporal dynamics of geometrid moth outbreaks in birch forest using MODIS-NDVI data

Defoliation caused by repeated outbreaks of cyclic geometrid moths is the most prominent natural disturbance factor in the northern-boreal birch forest. Evidence suggests that recent changes in outbreak distribution and duration can be attributed to climate warming. There is hence an immediate need for methods that can be applied to characterize the geographical distribution of outbreaks. Here we assess the reliability of MODIS (Moderate Resolution Imaging Spectroradiometer) 16-day NDVI data for generating time series of the distribution of defoliation caused by moths attacking birch forest in Fennoscandia. We do so by first establishing the relationship between ground measures of moth larval density and a defoliation score based on MODIS-NDVI. We then calibrate and validate a model with the MODIS-NDVI defoliation score as a classifier to discriminate between areas with and without visible defoliation as identified from orthophotos and provide two examples of application of the model. We found the MODIS defoliation score to be a valid proxy for larval density (R2 = 0.88–0.93) above a certain, low threshold (a defoliation score of ~ 5%). Areas with and without visible defoliation could be discriminated based on defoliation score with a substantial strength of agreement (max kappa = 0.736), and the resulting model was able to predict the proportion of area with visible defoliation in independent test areas with good reliability across the range of proportions. We conclude that satellite-derived defoliation patterns can be an invaluable tool for generating indirect population dynamical data that permits the development of targeted monitoring on relevant regional scales.