Assessing airborne laser scanning and spectral data for species-specific retention tree information

Abstract Key differences in biological legacy exist between clear-cutting and natural disturbances. One way to enhance similarity is by preserving structural features of old forests, such as retention trees, within harvested areas. Norwegian forest certification standards set by the Programme for the Endorsement of Forest Certification (PEFC) and the Forest Stewardship Council (FSC) require both the preservation and mapping of retention trees within harvested area for eventual reporting in a central database. This study, conducted in a managed forest in southeast Norway, evaluates the accuracy of retention tree identification, including density and volume predictions, using airborne laser scanning (ALS) data with low (2 pulses m –2 ) and high (approximately 100 pulses m –2 ) pulse densities, with and without spectral data. We also assess whether ex-situ reference data, such as diameter–height measurements from sample plots in similar forests or species annotations from aerial images, can fully or partially replace in-situ data collected within harvested stands.Three reference datasets were used, fully or partially: (1) 630 in-situ retention trees across 27 stands (for species classification and diameter at breast height (DBH) prediction), (2) 1064 ex-situ sample trees (for DBH prediction), and (3) 150 ex-situ annotated segments (for species classification). Using an individual tree segmentation approach with adaptative local maxima window size and regeneration height filtering, 65% of the in-situ retention trees were correctly identified, increasing to 74% when excluding snags. ALS at 2 pulses m –2 alone provided reliable total density and volume predictions, while spectral data improved species-specific accuracy. Species classifications remained consistent across data source (kappa = 0.556 for in-situ retention trees, 0.519 for ex-situ annotated segments), but DBH were underpredicted with ex-situ sample trees (RMSE = 9.4 cm, MSD = − 4.6 cm) compared to using 40 in-situ retention trees (RMSE = 8.8 cm, MSD = 0.2 cm). We recommend sampling approximately 40 in-situ retention trees to calibrate diameter-height models and using ex-situ annotated segments for species classification. This approach, based on low-density ALS and orthophoto datasets, meets the PEFC requirement to provide the locations of retention trees and may also support retrospective detection, thereby contributing to semi-automated certification reporting and virtual audits. In the eventuality of a publicly accessible database of measured retention trees were available, in-situ sampling for diameter-height model calibration could be omitted.