ing at the region of 33.5°N-34.5°N on the satellite transit route, we compare the aerosol vertical distribution characteristics during clean period, haze, dust, and pollution caused by fireworks based on CALIPSO satellite laser radar data. The results show that, in the sunny day, most of aerosols are clean continental aerosols at high altitude; in haze, most of aerosols are polluted continental aerosols whose backscatter and extinction are strong, and most of aerosol particles are spherical particles with small diameters; in dust, aerosols have a wide vertical distribution from the ground to the high altitude, and most of aerosol particles are non-spherical particles with big sizes; in pollution caused by fireworks, aerosol particles are the small size particles at low altitude, whose types are the polluted continental aerosol and the polluted dust aerosol. It is concluded that the vertical distributions of aerosol under different population types are different. We can use CALIPSO satellite laser radar data, together with meteorological element and HYSPLIT model to characterize the category of the atmospheric aerosol.

ing at the problem that the shallow machine learning algorithm commonly used in remote sensing image classification application cannot satisfy the classification accuracy in the current mass remote sensing image data environment, we propose a method to apply the fully convolution neural network to the remote sensing image classification. To reduce the loss of image feature map in the pooling process, we add the fusion of the pool layer and the deconvolution layer. To improve the reliability of fusion, we add the scale layer. To obtain finer edge classification results, considering the spatial correlation between pixels mean-shift clustering is used to obtain the spatial relationship of pixels. Classes of regional objects are determined by the maximum sum and the minimum variance of the regional pixel probabilities. Images of typical regions are chosen to carry out the classification experiments, and the classification method proposed in this paper is compared with those of the fully convolution neural network, support vector machine, and artificial neural network. The results show that the accuracy of the classification method proposed in this paper is obviously higher than that of the traditional machine learning methods.