Intra-annual Ips typographus outbreak monitoring using a multi-temporal GIS analysis based on hyperspectral and ALS data in the Bialowieza Forests

Disturbances caused by biotic and abiotic agents are inherent processes of forest ecosystems and often play a key role in forming the ecosystems themselves. However, they also may negatively affect wood production and alter ecosystem services. Thus, precise and up-to-date spatial information on invasion presence and dynamic is essential when planning protective and sanitary actions. Gathering such data is still challenging if large areas, with limited access, are affected. This study addresses this issue by suggesting a GIS-workflow based on high-resolution multi-temporal remote sensing data to describe the spread of a bark beetle outbreak by mapping dead trees on a single tree level and estimating timber losses associated with the infestations in the Polish part of the Bialowieza Forest (BF) in 2015. Along with field inventories, airborne hyperspectral data were collected in 2015 at three acquisition dates: 2-4 July, between August 24th and 27th and between 1st and 2nd October with a HySpex VNIR-1800 camera. Airborne laser scanning data were collected during 2-5 July 2015 using a Riegl LMS-Q680i scanner. The analysis consisted of three steps: (i) hyperspectral images were used to stratify the area into living vegetation and all other classes (including dead trees); (ii) a single tree delineation was performed and individual tree volumes were calculated based on a volume similar to height allometry; (iii) the delineated trees were assigned as either living trees or dead trees (snags). Based on these data bark beetle dynamics between the three time steps were analysed. The results show that the Ips typographus outbreak in BF in 2015 was very dynamic. The volume of dead standing spruce detected in subsequent dates was as follows: July - 503 410 m(3); August - 640 692 m(3); October - 737 978 m(3) New groups of dead trees in the flat landscape of Bialowieza forest appeared preferably but not only in the immediate vicinity of trees classified as dead in the previous period. New infestations were also identified over 100 m away from earlier infested trees. On the other hand, we could not identify any tendencies in terms of the spread direction which makes the prediction of the spread-dynamics challenging in our study area.