基于光谱特征的北京城区植被滞尘分布反演

植被叶片的滞尘量可以表征空气污染的程度, 分析城市植被滞尘的空间特征对于制定更为有效的空气污染控制政策具有重要的现实意义。基于北京市主城区采集的大叶黄杨、国槐、毛白杨和山桃等四种典型绿化植被叶片的滞尘量、光谱反射率和叶面积等数据，比较四种植被叶片滞尘前后的光谱曲线，进行窄波段与卫星波段滞尘前后叶片光谱反射率比值与滞尘量的相关分析。然后，分别建立相关性最大的卫星波段反射率和NDVI与滞尘量之间的回归模型, 选取拟合较好的模型反演北京城区植被的滞尘量分布，进而插值得到整个北京城区的尘埃分布。最后，根据高滞尘区域周围的土地覆盖和土地利用以及滞尘期间PM₁₀浓度的空间分布对反演的的合理性进行检验。结果表明：在780~1 300 nm波段，大叶黄杨、国槐、毛白杨和山桃四种植被的滞尘叶片反射率均明显低于干净叶片;窄波段反射率与滞尘量在520~650 nm波段和1 390~1 600 nm波段具有较高的相关性，相关系数的绝对值最高达到0.626;利用Landsat8的green波段和NDVI构建的滞尘反演模型，决定系数(R2)分别为0.446和0.465。NDVI模型反演的北京城区植被的滞尘量分布结果表明，北京城区滞尘含量呈现出北高南低，东高西低，中心城区高于郊区的空间分布格局。该研究通过高光谱和遥感影像数据反演滞尘量，可以为快速全面监测城市地区尘埃分布提供参考。

Retrieving Dustfall Distribution in Beij ing City Based on Ground Spectral Data and Remote Sensing

Dust-fall distribution of vegetation leaves can indicate the degree of air pollution;therefore the analysis of spatial char-acteristics of urban vegetation dust-fall has important practical significance for making more effective air pollution control policy. Based on the data of weight of dust,spectral reflectance and leaf area of Euonymus japonicus,Sophora japonica,poplar and da-vidiana collected in the main area of Beijing city,we compared the curve of spectrum of four plants “dust leaves”to “clean leav-es”;the correlation analysis between leaf spectral reflectance ratio (Dust/Clean)of narrow band and satellite band was processed with the weight of dust-fall respectively,with application of four plants leaf data.Then,we built the regression model of the satellite band reflectance and NDVI with dustfall weight respectively,and we used the best model to retrieve the dust-fall distri-bution of vegetation coverage area in Beij ing city,furthermore,we obtained the dust distribution of the whole Beij ing city through interpolation.Finally,we carried out the rationality verification of the result by the land cover and land use of the high dust region,as well as the average concentration of PM10 .The results showed that,dust leaves had an obviously lower reflec-tance than clean leaves in 780~1 300 nm which belonged to near-infrared bands;therewas a higher correlation between narrow band reflectance and dust-fall weight in 520~620 and 1 390~1 600 nm,up to -0.626;the coefficients of determination (R2 )of inversion models were respectively 0.446 and 0.465,which were constructed by green band and NDVI of Landsat8 with dust-fall weight.Using the model established with NDVI to retrieving the dust-fall distribution of Beij ing city,the results demonstrate that the distribution of dust-fall is high in north and low in south,high in east and low in west,high in downtown and low in the suburbs.This study offers a low-cost and effective method for investigating dust-fall distribution in urban area,and provides da-ta support to analysis sources of pollution for the environmental protection department.