We used fine-spatial resolution remotely sensed data combined with tree-ring parameters in order to assess and reconstruct disturbances. in mountain birch (Betula pubescens) forests caused by Epirrita autumnato (autumnal moth). Research was conducted in the area of Lake Tornetrisk in northern Sweden where we utilized five proxy parameters to detect insect outbreak events over the 19th and 20th centuries. Digital change detection was applied on three pairs of multi-temporal NDVI images from Landsat TM/ETM+ to detect significant reductions in the photosynthetic activity of forested areas during disturbed growing seasons. An image segmentation gap-fill procedure was developed in order to compensate missing scan lines in Landsat ETM+ "SLC-off" images. To account for a potential dependence of local outbreak levels on elevation, a digital elevation model was included in the defoliation recognition process. The resulting damage distribution map allowed for the assessment of outbreak intensity and distribution at the stand level and was combined with tree-ring data and historical documents to produce a multi-evidence outbreak detection. Defoliation events in the tree-ring data were recognized as significant deviations from temperature related growth.
Our outbreak detection scheme allowed for the reconstruction of nine major insect outbreaks over the past two centuries. The reconstruction proved reliable but only robust for severe defoliation events. Low-intensity incidents were not captured.