Work

Mapping large European aspens (Populus tremula L.) using national aerial imagery and a U-Net convolutional neural network

Year: 2025

Type: article

Abstract: Biodiversity supports boreal forest stability, with European aspen (Populus tremula L.) enhancing it by providing crucial habitats and resources for dependent species. Despite existing aspen remote sensing methods, there is no efficient method for mapping it over large areas to support forest management and conservation. This study developed a deep learning model to detect aspen from openly available national aerial imagery, evaluated its accuracy using field data, and compared the detection performance between leaf-on and leaf-off conditions. The study was conducted in three areas in southern Finland: Helsinki, Lohja, and Evo. We developed a fully convolutional semantic segmentation model for detecting aspens using a U-Net encoder-decoder architecture. The aerial imagery with four bands (RGB and NIR) at 0.5 m resolution was collected for leaf-on and leaf-off conditions. The training data for the model was gathered using visual observations of aspens in aerial imagery from 2010 to 2023, including 290 locations in Finland. We found notable differences in aspen detection accuracy between leaf-on and leaf-off conditions and size. The F1-score was higher in leaf-off (0.577) than leaf-on (0.463), with intersection over union (IoU) values of 0.280 and 0.256 for leaf-off and leaf-on, respectively. The diameter at breast height (dbh) of the detected aspens was similar for both conditions. Moreover, the detection accuracy improved with increasing aspen dbh, with F1-scores reaching 0.663 (leaf-off) and 0.551 (leaf-on) for >20 cm, and 0.710 (leaf-off) and 0.594 (leaf-on) for >30 cm. The developed model reasonably locates aspen distribution and abundance, assisting forest managers make informed management decisions.

Source: Remote Sensing Applications Society and Environment

DOI: 10.1016/j.rsase.2025.101755

Authors: Anwarul Islam Chowdhury, Einari Heinaro, Topi Tanhuanpää, Anis Ur Rahman, Samuli Junttila

Institutions: University of Helsinki, University of Eastern Finland

Language: English

Topic: Remote-Sensing Image Classification

Subfield: Media Technology

Field: Engineering

Domain: Physical Sciences

SDG: Life on land

Open Access status: hybrid

Estimated APC paid: $2,640