地表植被景观对PM2.5浓度空间分布的影响研究

我国当前城市日益频发的雾霾问题引发公众广泛关注，PM_2.5被认为是雾霾的主要成因。研究认为，在某一区域短时间尺度上（如日），PM_2.5浓度主要受气象条件影响。但在较长时间尺度上（如季，年），由于气象条件基本相似，则PM_2.5浓度主要受土地利用特别是地表植被景观的影响。如何耦合地表植被景观格局与PM_2.5浓度信息，定量分析其影响是当前相关科学研究的一个难点，需要引入新思路。首先基于季节气象条件基本相似的科学假设，采用土地利用回归模型分四季高精度模拟PM_2.5浓度空间分布。其次，采用像元二分模型分四季估算研究区植被覆盖度。在此基础上采用随机抽样法通过统计回归模型耦合植被覆盖度与PM_2.5空间分布，定量研究植被覆盖度对PM_2.5分布影响及其尺度效应。研究结果表明：1）植被覆盖度与PM_2.5浓度在本研究选择的空间尺度上，都显著负相关，说明植被覆盖度对PM_2.5具有显著影响；同一个季节不同尺度上，以及不同季节同一尺度上的植被覆盖度对PM_2.5浓度的影响存在一定差异。2）植被覆盖度对PM_2.5浓度的影响方式比较复杂，不同的季节的表现方式不同，总体来说PM_2.5浓度与植被覆盖度曲线回归模型的拟合度高于线性回归模型，说明植被覆盖度对PM_2.5的影响具有非线性特征。3）不同的PM_2.5浓度水平下，植被覆盖度对PM_2.5浓度的影响程度存在差异。PM_2.5浓度越高，植被覆盖度对其浓度的影响越明显。本研究提出的区域尺度耦合地表植被覆盖与PM_2.5浓度的思路与方法，有效的揭示了植被覆盖度对PM_2.5浓度分布的影响方式与尺度效应，为通过优化城市植被缓解大气污染提供一定参考。

The impacts of vegetation landscape on PM2.5 spatial distribution

The frequently breaking out fog and haze has attracted widespread public attention in China recently. PM_2.5 is considered to be main cause of it. Recent studies show that PM_2.5 is mainly controlled by the meteorological conditions in short temporal scale, for example, daily, but strongly affected by land cover, especially vegetation in long temporal scale such as yearly, seasonal. It is a challenge for landscape ecology to couple vegetation landscape pattern and PM_2.5 spatial information in order to quantitatively analyze the impacts. New idea and method are required to cope with it. Based on the assumption of a stable meteorological conditions within a season, this paper first used the land use regression (LUR) model to precisely simulate the spatial distribution of seasonal PM_2.5 concentration. Secondly, based on the pixel dichotomy model, the vegetation coverage of the study area was estimated for four seasons and the spatial distribution was determined. Finally, the impacts of vegetation landscape on PM_2.5 and the scale effects were discovered by means of pixel random sampling and regression models. The results showed as follows. 1) There existed a significant negative correlation between vegetation coverage and PM_2.5 concentration in the spatial scales adopted in this study. The influences not only varied from season to season, but also changed at different scales within the same season. 2) The forms of the effects of vegetation coverage on concentration were complicated. In general, the curve regression models were better than the linear ones, indicating that the relationship between vegetation coverage and PM_2.5 concentration was non-linear. 3) The higher the PM_2.5 concentration, the sharper the constructed curve model, and the stronger influence the vegetation coverage had on PM_2.5 concentration. This study puts forward a new idea to couple vegetation coverage and PM_2.5 to discover their relationship and scale effects in regional scale. It is expected to provide a reference for mitigating atmospheric pollution by optimizing urban vegetation landscape.