Kaukokartoitus mahdollistaa tarkan tiedon tuottamisen metsäluonnon monimuotoisuutta ilmentävistä piirteistä

Abstrakti

Up-to-date knowledge of key ecological features that maintain boreal forest biodiversity is essential part of sustainable forest management and conservation measures. However, there is only a limited amount of spatial data available, as the detection of ecologically significant elements using remote sensing is challenging due to their low frequency, scattered occurrence, small size and/or location in the field layer. We studied tree-level discrimination of the keystone species aspen (Populus tremula L.) from other common tree species and detection of deadwood utilizing airborne high-resolution RGB images, multispectral and hyperspectral data and airborne laser scanning (ALS) data. We also tested the predictive accuracy of remote sensing parameters in predicting the aspen epiphytic lichen communities. Our results demonstrate that high-resolution remote sensing data together with machine learning algorithms provide new possibilities for mapping the occurrence and spatial distribution of key ecological features and producing proxies of biodiversity. It should be noted that field work is also needed for model training and evaluation. Moreover, resources and collaboration of different organizations are needed for biodiversity assessment and monitoring over large geographical areas.