基于激光雷达观测的大气边界层自动识别局部最优点算法

发布时间：2018-07-06 13:57

本文选题：激光雷达 + 污染输入　；参考：《光谱学与光谱分析》2017年02期

【摘要】：大气边界层高度是影响近地面大气物理运动的主要因素,同时也是影响地面污染物浓度的一个重要因素。地基激光雷达可以对大气气溶胶的垂直分布进行连续稳定的监测,应用激光雷达技术对大气边界层进行连续观测可以为环境监测与预报提供指导性的动态信息。针对存在残留层以及外来污染物输入情况时边界层高度变化检测的可靠性及计算效率问题,结合梯度法的物理意义与激光雷达时序图的图形图像学特征,提出了一种基于时空邻近度的边界层局部最优点识别算法。以江苏省无锡市新区偏振米散射激光雷达太湖观测站点的气溶胶垂直观测数据为例,通过对2012年底两次污染事件进行观测分析,分别使用梯度法和局部最优点法进行大气边界层高度的自动识别。实验结果表明,在静稳状态和污染混入后的情况下,梯度法与局部最优点识别法的结果较为接近,但梯度法在处理污染混入状态以及存在残留层的情况下误判率较高。基于时空邻近度的局部最优点算法通过对垂直特征值以及水平相关性的控制,有效地消除了在弱信号、噪声信号、低云以及存在残留层和外来污染等情况下导致的计算机误判现象,在减小算法时间复杂度的同时在计算机自动识别结果具有更高的稳定性,弥补了梯度法在自动化运行中的识别精度与计算效率的不足。
[Abstract]:The height of the atmospheric boundary layer is the main factor affecting the physical movement of the atmosphere near the ground, and it is also an important factor affecting the concentration of pollutants on the ground. Ground-based lidar can continuously and stably monitor the vertical distribution of atmospheric aerosols. Continuous observation of atmospheric boundary layer by using lidar technology can provide guiding dynamic information for environmental monitoring and prediction. Aiming at the reliability and computational efficiency of detecting the height change of the boundary layer in the presence of residual layer and the input of foreign pollutants, the paper combines the physical meaning of gradient method and the graphic and image characteristics of the lidar sequence diagram. A local optimal recognition algorithm for boundary layer based on temporal and spatial proximity is proposed. Taking the aerosol vertical observation data from Taihu Lake observation station of polarization meter scattering lidar in Wuxi City, Jiangsu Province as an example, two pollution events were observed and analyzed at the end of 2012. The gradient method and the local optimal point method are used to identify the height of the atmospheric boundary layer. The experimental results show that the gradient method is close to the local optimum recognition method in the case of static and stable state and after the contamination is mixed, but the gradient method has a higher misjudgment rate in the case of dealing with the mixed state of pollution and the existence of residual layer. By controlling vertical eigenvalues and horizontal correlations, the local optimal algorithm based on spatio-temporal proximity effectively eliminates weak signals and noise signals. In the case of low cloud and the existence of residual layer and external pollution, the computer misjudgment phenomenon can reduce the time complexity of the algorithm, and at the same time, the computer automatic recognition results have higher stability. It makes up for the deficiency of recognition accuracy and calculation efficiency of gradient method in automatic operation.
【作者单位】：中国矿业大学环境与测绘学院;国土环境与灾害监测国家测绘地理信息局重点实验室;江苏省资源环境信息工程重点实验室;无锡中科光电技术有限公司;
【基金】：中央高校基本科研业务费专项资金项目(2014QNA32)资助
【分类号】：TN958.98

【相似文献】