Historical insect disturbance maps from 1985 onwards for Canadian forests derived using earth observation data

Despite the major impact of insect outbreaks on the Canadian boreal forest and its significance in carbon monitoring, current monitoring efforts primarily rely on costly and subjective aerial survey interpretations. While satellite remote sensing has been widely used to map wildfire and harvesting disturbances, no consistent, long-term dataset exists for severe canopy loss events in coniferous forests. This paper presents the development and evaluation of annual maps of boreal forest insect pest severe disturbances in Canada from 1985 to 2024. We introduce a methodology that leverages Landsat imagery with a 30 m spatial resolution to provide a standardized, long-term record of severe pest-related defoliation. The overall prediction accuracy between the aggregated moderate and severe pest and non-pest classes was evaluated as 90%. This historical dataset offers valuable insights for forest ecology and disturbance monitoring and research, forest carbon modeling, and forest management.