Estimation Of Vegetation Dust Deposition Distribution Based On Multispectral Image

With the rapid development of urban areas,the improvement of urban transportation infrastructure and the growth of traffic volume are the main reasons for the deterioration of urban air quality resulting from the increase of atmospheric particulate matter concentration.Vegetation plays an important role in the self-regulation function of ecosystem.Using remote sensing technology to monitor and analyze vegetation dust retention can quickly and effectively measure the regional environmental air quality from the perspective of urban green space allocation.In this paper,five common urban vegetation species were selected as the study objects in Xuhui and Minhang districts of Shanghai,China.To monitor the ground hyperspectral data and dust deposition data of leaves,then,combined with remote sensing image data（Sentinel-2）,based on vegetation index,explored the relationship between vegetation dust deposition and remote sensing satellite spectral information,constructed the inversion models of regional vegetation dust deposition amount using Partial Least Squares Regression（PLSR）and BP neural network regression,estimated the spatial distribution of vegetation dust deposition,evaluated the total vegetation dust deposition in the study area.The main results are as follows:（1）NDVI,EVI,OSAVI and SAVI can be used as the VI of remote sensing satellite which are more sensitive to dust deposition,and can be used to construct the inversion model of dust deposition amount.In the correlation matrix of dust deposition capacity by vegetation and remote sensing satellite spectral vegetation index,it was observed that the correlation coefficient（R）between vegetation index and dust deposition amount was basically good,NDVI（0.86）and EVI（0.85）were higher,followed by OSAVI（0.76）and SAVI（0.69）,and NLI（0.36）was the lowest.（2）The multivariate inversion model based on BP neural network regression has a better ability to predict the regional vegetation dust deposition.SAVI,OSAVI,EVI and NDVI as independent variables to build multivariate model accuracy is higher than NDVI as independent variables to build Unigram.Inversion model based on BP neural network regression to build better compared with PLSR method,and verified the fitting accuracy of the model with 20 validation samples,the results showed that the determination coefficient（R²）reached a significant level in the two multivariate models,and there was little difference in the fitting accuracy.However,the model established by BP neural network regression method that R² verified was higher as0.8632.（3）The spatial distribution of vegetation dust retention was positively correlated with the area of green space on the ground and its continuity.Based on the data from the Sentinel-2 and the optimal inversion model constructed,the Kriging interpolation method was used to quantify the spatial distribution of vegetation dust deposition amount in the study area.The result showed that on August 17,2017,the dust deposition amount of vegetation in the study area was generally higher in the central urban area and lower in the suburban area.The dust deposition amount in the south of Xuhui District of Shanghai was relatively high,with the highest average value of 41gm^-2.The dust deposition amount in the east of Huangpu River and the southwest corner of Minhang District was relatively high,with the highest average value of 49gm^-2 and 43 gm^-2,respectively.Combined with Google Earth and vegetation information extracted,it is found that there is a continuous green spatial distribution in these areas.（4）According to the Spatiotemporal dry deposition model to estimate the PM_2.5remove amount of vegetation in the study area in August 2017.The total area covered by urban green space in Xuhui District and Minhang District of Shanghai was about89.3357 km²,the total amount of PM_2.5 removed by vegetation was about 345.4902 t,and the average removal amount per unit leaf area was 3.8673 gm^-2mo^-1.The dust deposition rate of vegetation was higher in the south of Xuhui District and along the east of Huangpu River in Minhang District,with the highest values of 4.6270gm^-2mo^-1 and 6.2183 gm^-2mo^-1,respectively.In addition,due to the concentration of commercial land in the central urban area,the dust deposition effect of vegetation was weakened.Therefore,proper allocation of green space in these areas is recommended.