Forest attribute maps: a support for small area estimation of forest disturbances

Abstract Key Message Forest attribute maps generated using national forest inventory and remote sensing data can be used to quantify forest attributes (i.e., growing stock volume, basal area) affected by disturbance events (like bark-beetle induced damage) through small area estimation techniques. Our results showed that mature stands were more sensitive to bark-beetle infestation than younger ones. Context Recurrent drought episodes in Europe weakened trees and promoted biotic disturbances such as bark-beetle outbreaks, heavily affecting spruce ( Picea abies (L.) H. Karst.) and fir stands (Abies alba Mill.). While remote sensing data make it possible to assess and monitor the spatial extent of outbreaks, the assessment of the actual impact on forest attributes is still lacking. Aims We propose a small area estimation approach to assess the amount of forest attributes impacted in outbreak areas during the period 2021–2022 using forest attribute maps. Methods We produced forest attribute maps using National Forest Inventory (NFI) plots, and auxiliary data from Light Detection and Ranging (LiDAR) and Sentinel-2, acquired in 2021. The maps were generated using a general k-nearest neighbour model for all forest species (general model) and a model specific to spruce and fir only (specific model). Small estimation domains, defined as bark-beetle outbreak areas, were obtained from an independent source using time series of Sentinel-2 data (2018–2022). Small area estimations were generated using bootstrapped model-predictions with reliability assessment. Results In total, 1137 ha of forest area was affected by bark-beetle outbreaks during 2021–2022, corresponding to 494,000 m 3 (general model) and 505,000 m 3 (specific model) of growing stock volume, respectively. The associated errors were 2.2% for the general model and 2.6% for the specific model. In the specific model, the estimated mean growing stock volume in outbreak areas for 2021 and 2022 was 441.5 m 3 ha⁻ 1 , which was 64.5 m 3 ha⁻ 1 higher than the estimated mean volume of spruce and fir for the whole area (377 m 3 ha⁻ 1 ). This suggests that mature stands were more sensitive to infestation. Conclusion Regular updates to forest attribute maps, possibly every third year using photogrammetric methods, could improve monitoring of forest attributes impacted by disturbances such as drought and bark-beetle outbreaks and contribute to forest decision making and planning.