Environmental Science Programs, School of Science, Hong Kong University of Science and Technology, Clear Water Bay, Hong Kong, China.
Atmospheric Research Center, HKUST Fok Ying Tung Graduate School, Nansha IT Park, Guangzhou 511458, China.
Sci Total Environ. 2015 Feb 15;506-507:527-37. doi: 10.1016/j.scitotenv.2014.11.072. Epub 2014 Nov 26.
The Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model is widely used to generate backward trajectories in given starting locations. However, differences exist between trajectories generated from the model with different input datasets. In this study, backward trajectories in Hong Kong in the entire year of 2011 are derived by HYSPLIT model. Two sets of Global Data Assimilation System (GDAS) output data associated with different horizontal and vertical resolutions (GDAS1 and GDAS0P5) are used as drivers in an attempt to quantify the differences between the results and discover the underlying reasons responsible for discrepancy. The results reveal that the significant differences between back trajectories generated from the two GDAS datasets can be mainly attributed to different vertical velocity calculation methods due to the absence of vertical velocity in GDAS0P5 dataset. The HYSPLIT trajectories are also sensitive to the horizontal and vertical resolutions of the input meteorological data, but to lesser extents. Results of cluster analysis indicate that when the air mass is from the north, northeast, or west with a long-to-medium range, the HYSPLIT backward trajectories are sensitive to the vertical advection calculation method and data resolution, whereas when the air mass is from the south or southwest with a long range, the trajectories are more likely to remain unchanged with the shifting of vertical velocity or data resolution. By comparing the vertical velocities with the observations and the performance in retrieving PM contributions from different directions, we conclude that GDAS1 dataset is more plausible in backward trajectory analysis in the Pearl River Delta.
混合单粒子拉格朗日综合轨迹（HYSPLIT）模型被广泛用于在给定起始位置生成回溯轨迹。然而，不同输入数据集生成的轨迹之间存在差异。本研究使用 HYSPLIT 模型生成了香港 2011 年全年的回溯轨迹。使用两套与不同水平和垂直分辨率相关的全球数据同化系统（GDAS）输出数据（GDAS1 和 GDAS0P5）作为驱动，试图量化结果之间的差异并找出导致差异的根本原因。结果表明，由于 GDAS0P5 数据集缺少垂直速度，因此由两个 GDAS 数据集生成的回溯轨迹之间存在显著差异，主要归因于不同的垂直速度计算方法。HYSPLIT 轨迹也对输入气象数据的水平和垂直分辨率敏感，但程度较小。聚类分析结果表明，当气团来自北方、东北方或西方，且距离较长或中等时，HYSPLIT 回溯轨迹对垂直平流计算方法和数据分辨率较为敏感，而当气团来自南方或西南方且距离较远时，轨迹更可能随着垂直速度或数据分辨率的变化而保持不变。通过比较垂直速度与观测值以及从不同方向检索 PM 贡献的性能，我们得出结论，GDAS1 数据集在珠江三角洲的回溯轨迹分析中更为合理。
