基于无人机影像的无瓣海桑单木提取与地上生物量估算

无瓣海桑(Sonneratia apetala)是中国红树林恢复种植最早引进的优质红树树种，其生产力在红树林群落中处于较高水平，具有显著的高生物量和能量积累。然而，由于红树林群落冠层密集、结构复杂，精确描绘无瓣海桑的单木树冠存在极大挑战性。传统的卫星遥感侧重于区域或更大尺度监测需求，而新兴的低空无人机遥感在更精细尺度的红树林生态监测中具有显著优势。以广东省珠海市淇澳岛红树林自然保护区为研究区，利用消费级无人机影像生成的冠层高度模型(Canopy Height Model, CHM)和种子区域生长(Seed Region Growing, SRG)算法进行无瓣海桑单木树冠提取，并建立基于地面调查数据获取的树高和胸径两者之间的回归关系，以优化无瓣海桑地上生物量异速生长方程，进而实现研究区单木尺度的无瓣海桑地上生物量估算。结果表明：基于无人机影像可以有效提取无瓣海桑单木树冠，其提取精度达到67%；验证了树高和胸径之间较高的相关性，提出了基于树高的无瓣海桑地上生物量异速生长方程；研究区无瓣海桑平均地上生物量的范围为2.99～247.24 t/hm²，平均值为92.14 t...

Individual Tree Crown Delineation and Aboveground Biomass Estimation of Sonneratia apetala Based on Unmanned Aerial Vehicle Remote Sensing Images

Accurate mangrove biomass measurement is necessary for the management and protection of mangrove ecosystems. Sonneratia apetala was the first high-quality mangrove species to be introduced for mangrove restoration in China. Compared with other mangrove species, Sonneratia apetala has higher productivity and can store large amounts of carbon in its living biomass. However, accurate depiction of the single-wood canopy of Sonneratia apetala is challenging because of its high clumping density and intricate canopy structure. While traditional satellite remote sensing focuses on regional or larger-scale monitoring needs, the newly emerged Unmanned Aerial Vehicle (UAV) remote sensing has significant advantages for monitoring mangroves at finer scales. However, few studies have used UAV data for mangrove biomass analyses. In this study, we successfully used consumer-grade UAV data to estimate the height and aboveground biomass (AGB) of Sonneratia apetala on Qi'ao Island, Zhuhai, Guangdong Province. We used a variable window filter algorithm to detect the treetops. Individual tree canopy segmentation was performed using the seed region-growing algorithm. Additionally, we constructed a regression equation for height (H) and diameter at breast height (DBH) of Sonneratia apetala in the study area and optimized the traditional allometric equation. Finally, mangrove AGB was estimated at the tree level using the optimized allometric equation, and the results indicated that the AGB of Sonneratia apetala could be accurately extracted using UAV images. The accuracy of the tree delineation was 67%, the correlation between H and DBH was DBH = 2.2726H-6.4415, and the correlation coefficient R² was 0.8713. The aboveground mass of a single wood of Sonneratia apetala in the study area ranged from 29.60 to 388.44 kg, with a mean value of 145.72 kg and a total aboveground mass of 368.97 t. Partial spatial clustering was observed in the distribution of the aboveground mass of Sonneratia apetala in the study area, with a Moran's I index value of 0.594. The aboveground mass of Sonneratia apetala at the edge of the area and in the window part of the area was found to be smaller, mainly for two possible reasons. First, the natural death of Sonneratia apetala in part of the study area created a window, but its surrounding seedlings were still in the biomass accumulation stage. Second, the aboveground mass of Sonneratia apetala at the edges of the study area is often lost due to its vulnerable nature and anthropogenic factors. The average number of Sonneratia apetala in each quadrat was 6.3 and the AGB in the study area ranged from 2.99 to 247.24 t/hm², with a mean value of 92.14 t/hm². The estimated AGB based on UAV data was consistently lower than that generated from field data. The methods described in this study offer the possibility of easily repeatable, low-cost UAV surveys, providing a faster and more economical approach for monitoring mangrove forests than traditional ground surveys. These results may assist in decision-making regarding ecological monitoring, resource use, mangrove introduction, and scientific advancement of mangroves in China.