Prediction of forest canopy fuel parameters in managed boreal forests using multispectral and unispectral airborne laser scanning data and aerial images. Issue 1 (1st January 2020)

• Record Type: Journal Article
• Authors: Maltamo, Matti; Räty, J.; Korhonen, L.; Kotivuori, E.; Kukkonen, M.; Peltola, H.; Kangas, J.; Packalen, P.
• Is Part Of: European journal of remote sensing. Volume 53:Issue 1(2020)
• Page Start: 245
• Page End: 257
• Publication Date: 2020-01-01
• Subjects: Forest fire -- forest structure -- boreal forests -- fuel models -- forest fuel parameters -- LiDAR; Remote sensing -- Periodicals; Remote sensing; Electronic journals; Periodicals; 621.3678
• DOI: 10.1080/22797254.2020.1816142
• Languages: English
• ISSNs: 2279-7254

Abstract

ABSTRACT: This study evaluated the suitability of different airborne laser scanning (ALS) datasets for the prediction of forest canopy fuel parameters in managed boreal forests in Finland. The ALS data alternatives were leaf-off and leaf-on unispectral and leaf-on multispectral data, alone and combined with aerial images. Canopy fuel weight, canopy base height, biomass of living and dead trees, and height and biomass of the understory tree layer were predicted using regression analysis. The considered categorical forest parameters were dominant tree species, site fertility and vertical forest structure layers. The canopy fuel weight was modeled based on crown biomass with an RMSE% value of 20–30%. The canopy base heights were predicted separately for pine and spruce stands with satisfactory results the RMSE% values being 9–10% and 15–17%, respectively. Following the initial classification of the existence of an understory layer (with kappa-values of 0.47–0.53), the prediction of understory height performed well (RMSE% 20–25%) but the understory biomass was predicted with larger RMSE% values (about 60–70%). Site fertility was classified with kappa-values of 0.5–0.6. The most accurate results were obtained using multispectral ALS data, although the differences between the datasets were minor.