Assessing spatiotemporal variations of forest carbon density using bi-temporal discrete aerial laser scanning data in Chinese boreal forests

Abstract

Assessing the changes in forest carbon stocks over time is critical for monitoring carbon dynamics, estimating the balance between carbon uptake and release from forests, and providing key insights into climate change mitigation. In this study, we quantitatively characterized spatiotemporal variations in aboveground carbon density (ACD) in boreal natural forests in the Greater Khingan Mountains (GKM) region using bi-temporal discrete aerial laser scanning (ALS) data acquired in 2012 and 2016. Moreover, we evaluated the transferability of the proposed design model using forest field plot data and produced a wall-to-wall map of ACD changes for the entire study area from 2012 to 2016 ​at a grid size of 30 ​m. In addition, we investigated the relationships between carbon dynamics and the dominant tree species, age groups, and topography of undisturbed forested areas to better understand ACD variations by employing heterogeneous forest canopy structural characteristics. The results showed that the performance of the temporally transferable model (R2 ​= ​0.87, rRMSE ​= ​18.25%), which included stable variables, was statistically equivalent to that obtained from the model fitted directly by the 2016 field plots (R2 ​= ​0.87, rRMSE ​= ​17.47%). The average rate of change in carbon sequestration across the entire study region was 1.35 ​Mg·ha−1·year−1 based on the changes in ALS-based ACD values over the course of four years. The relative change rates of ACD decreased as the elevation increased, with the highest and lowest ACD growth rates occurring in the middle-aged and mature forest stands, respectively. The Gini coefficient, which represents forest canopy surface structure heterogeneity, is sensitive to carbon dynamics and is a reliable predictor of the relative change rate of ACD. This study demonstrated the applicability of bi-temporal ALS for predicting forest carbon dynamics and fine-scale spatial change patterns. Our research contributed to a better understanding of the influence of remote sensing-derived environmental variables on forest carbon dynamic patterns and the development of context-specific management approaches to increase forest carbon stocks.