Accurate and reliable information on tree species composition and distribution is crucial in operational and sustainable forest management. Developing a high-precision tree species map based on time series satellite data is an effective and cost-efficient approach. However, we do not quantitatively know how the time scale of data acquisitions contributes to complex tree species mapping. This study aimed to produce a detailed tree species map in a typical forest zone of the Changbai Mountains by incorporating Sentinel-2 images, topography data, and machine learning algorithms. We focused on exploring the effects of the three-year time series of Sentinel-2 within monthly, seasonal, and yearly time scales on the classification of ten dominant tree species. A random forest (RF) and support vector machine (SVM) were compared and employed to map continuous tree species. The results showed that classification with monthly datasets (overall accuracy (OA): 83.38–87.45%) outperformed that with seasonal and yearly datasets (OA:72.38–85.91%), and the RF (OA: 81.70–87.45%) was better than the SVM (OA: 72.38–83.38%) at processing the same datasets. Short-wave infrared, the normalized vegetation index, and elevation were the most important variables for tree species classification. The highest classification accuracy of 87.45% was achieved by combining RF, monthly datasets, and topography information. In terms of single species’ accuracy, the F1 scores of the ten tree species ranged from 62.99% (Manchurian ash) to 97.04% (Mongolian Oak), and eight of them obtained high F1 scores greater than 87%. This study confirmed that monthly Sentinel-2 datasets, topography data, and machine learning algorithms have great potential for accurate tree species mapping in mountainous regions.

Understanding accurate and continuous forest dynamics is of key importance for forest protection and management in the Greater Khingan Mountains (GKM). There has been a lack of finely captured and long-term information on forest disturbance and recovery since the mega-fire of 1987 which may limit the scientific assessment of the GKM’s vegetation conditions. Therefore, we proposed a rapid and robust approach to track the dynamics of forest disturbance and recovery from 1987 to 2021 using Landsat time series, LandTrendr, and random forests (RF) models. Furthermore, we qualified the spatial characteristics of forest changes in terms of burn severity, topography, and distances from roads and settlements. Our results revealed that the integrated method of LandTrendr and RF is well adapted to track forest dynamics in the GKM, with an overall accuracy of 0.86. From 1987 to 2021, forests in the GKM showed a recovery trend with a net increase of more than 4.72 × 10⁴ ha. Over 90% of disturbances occurred between 1987 and 2010 and over 75% of recovery occurred between 1987 and 1988. Mildly burned areas accounted for 51% of forest disturbance and severely burned areas contributed to 45% of forest recovery. Forest changes tended to occur in zones with elevations of 400–650 m, slopes of less than 9°, and within 6 km of roads and 24 km of settlements. Temporal trends of forest disturbance and recovery were mainly explained by the implementation timelines of major forestry policies. Our results provide high-resolution and time-series information on forest disturbance and recovery in the GKM which could support scientific decisions on forest management and sustainable utilization.

目的多光谱遥感图像林地变化检测是生态监测领域的重要研究方向之一。针对双时相图像中成像条件差异引起的伪变化，以及复杂地物变化在尺度与形态上的多样性问题，设计了一种频域增强分层交互网络，以提升复杂场景下林地变化检测的准确性与鲁棒性。方法首先，构建频域特征增强模块，在频域对双时相特征进行增强，以抑制由成像条件不一致引起的风格差异等伪变化干扰。其次，设计跨时相相关引导交互模块，通过对双时相特征的充分交互与相关性引导，增强网络对局部细微变化区域的识别能力。最后，通过空-谱耦合交互模块，对不同网络深度的变化特征进行整合，融合浅层细粒度结构信息与深层语义信息，实现对多尺度、多形态林地变化对象的联合感知。结果实验在3个林地变化检测数据集上与Spectral-Former、CSANet（cross-temporal interaction symmetric attention network）、TriTF（triplet Transformer framework）、DIEFEN（differential information-enhanced feature exchange network）、FrFTML（fractional Fourier transform domain decoupling-based adaptive weighted metric learning network）和AIWSEN（Adaptive information weighting and synchronized enhancement network）等先进方法进行了对比分析。在R1数据集中，所提出方法的Kappa系数达到0.8102，相比性能次优方法提升1.09%；在R2数据集中，Kappa系数提升2.62%；在R3数据集中，Kappa系数提升1.74%。同时，在三个数据集上的消融实验进一步验证了所提模块的有效性。结论本文所提方法能有效缓解双时相风格差异引起的伪变化干扰，并提升对细微变化与多尺度变化对象的检测能力，从而获得更为准确、稳定的林地变化检测结果。