Data fusion of atmospheric ozone remote sensing Lidar according to deep learning

This article aims to realize the combined application of satellite remote sensing data and Lidar data fusion of atmospheric ozone in the ozone distribution prediction in western China. Firstly, the ozone remote sensing data and Lidar vertical monitoring data detected by ozone monitoring instruments are collected from various provinces in western China; the recurrent neural network (RNN) model based on the gated recurrent unit and the convolutional neural network (CNN) model based on time series are constructed and integrated to obtain the RNN-CNN model for ozone concentration prediction, and the prediction performance is compared with that of the RNN and CNN model. Satellite remote sensing data show that the total ozone concentration in western China has obvious spatial distribution characteristics, and can be elevated with the increase in latitude; the ozone concentration in the north is higher and the ozone concentration in the south is lower in different seasons in western China, and it in autumn and summer is lower than that in spring and winter generally. The hour-by-hour change of ozone concentration shows that the ozone concentration in different seasons reaches a trough around 9-10 o'clock, and reaches a peak around 16-17 o'clock. When the RNN-CNN model is applied to predict the hourly change of ozone concentration within a day, it can achieve the best root-mean-squared error (18.13), mean absolute error (11.64), and index of agreement (0.973). The monitoring results of Lidar data at a certain site are compared and analyzed, and the results show that the fitness is the best between the predicted value of ozone concentration obtained by the RNN-CNN model and the actually measured value. This research provides the possibility for the intelligent prediction of atmospheric ozone concentration distribution and has reliable theoretical value and practical significance.