基于下垫面年地表温差的武汉市PM10空间分布的估计

ESTIMATES OF THE SPATIAL DISTRIBUTION OF PM_(10) BASED ON THE SURFACE TEMPERATURE DIFFERENCE OF UNDERLYING SURFACE IN WUHAN,CHINA

城市可吸入颗粒物(PM_(10))某一时刻或短期的空间分布,主要受气象条件控制,而一年或多年平均分布则主要取决于排放源。这些排放源与城市交通道路、工业区、城市建成区和开发区等的下垫面分布密切相关,而年地表温差可以综合反映下垫面的这些特性,所以可以利用这种相关关系,建立模型来估计年平均PM_(10)的空间分布。以武汉市为例,首先利用Landsat 8热红外遥感数据反演出2013年和2014年夏天和冬天的地表温度,计算出地表温差值;然后,根据影响随距离衰减的地学原理,利用反距离加权法(IDW),得到任意像元处年地表温差加权值,并与地面实测的2013年和2014年PM_(10)年均值做一元线性回归,通过精度对比寻找到最佳年地表温差加权值,并得到空间分布估计模型,其拟合优度R2达到0.655和0.752;最后利用该模型得到武汉市2013年和2014年PM_(10)年均值空间分布图。结果表明,武汉PM_(10)年均值浓度高值区主要集中于主城区,郊区部分人口相对集中的区域PM_(10)也较高,低值区分布在郊区乡镇、偏远山区以及有大型水体的地方。由于新方法充分考虑了下垫面的影响,与克里金内插相比,更能精细地刻画和反映PM_(10)的分布特征和规律,而且简单有效,有一定的应用价值。

The short-term spatial distribution of City ＇s respiratable particulate matter（PM_（10）） is mainly controlled by weather conditions,while the distribution over one year or longer is mainly dependdent on the sources of emissions. These sources of emissions are closely related to the distribution of urban traffic,industrial zones,urban built-up areas,and developing zones. While surface temperature difference within a year can integrate these features of underlying surface. So we can use this correlation and set up a model to estimate the spatial distribution of annual average PM10. Taking Wuhan as an example,we firstly used Landsat 8 thermal remote sensing data to inverse the summer and winter surface temperature of 2013 and 2014,and calculated the surface temperature difference of every year. Then, according to the principles of impact attenuation with distance,we used inverse distance weighting（IDW） to get the weighted value of every year surface temperature difference. After that, we formulated a linear regression between the weighted value of every year surface temperature difference and the PM10 annual average of 2013 and 2014,with an aim to find the best weighted value by comparing the accuracy,and to get the spatial distribution of the estimated model; its R2 reached 0. 655 and 0. 752,respectively. Finally,we got the spatial distribution of annual average PM10 in 2013 and 2014 based on the model. The spatial distribution of PM10 showed that the high value of 2013 and 2014 PM10 annual average mainly concentrated in the main city compared to suburban area. Low value is located mainly in the outskirts of towns,remote and mountainous areas or large bodies of water. Compared with the Kriging interpolation,the new method takes into account the influence of the underlying surface, so it can finely reflect the distribution characteristics and laws of PM10. It is more simple and effective.