Abstract
The applications of novel deep learning techniques in atmospheric science are rising quickly. Here we build a hybrid deep learning (DL) model (hyDL-CO), based on convolutional neural networks (CNN) and long short-term memory (LSTM) neural networks to provide a comparative analysis between DL and Kalman Filter (KF) to predict carbon monoxide (CO) concentrations in China in 2015–2020. We find the performance of DL model is better than KF in the training period (2015–2018): the mean bias and correlation coefficients are 9.6 ppb and 0.98 over E. China, and −12.5 ppb and 0.96 over grids with independent observations. By contrast, the assimilated CO concentrations by KF exhibit comparable correlation coefficients but larger negative biases. Furthermore, DL model demonstrates good temporal extensibility: the mean bias and correlation coefficients are 95.7 ppb and 0.93 over E. China, and 81.0 ppb and 0.91 over grids with independent observations in 2019–2020, while CO observations are not fed into the DL model as an input variable. Despite these advantages, our analysis indicates a noticeable underestimation of CO concentrations at extreme pollution events in the DL model. This work demonstrates the advantages and disadvantages of DL models to predict atmospheric compositions in respective to traditional data assimilation, which is helpful for better applications of this novel technique in future studies.
Highlights
Our analysis indicates a noticeable underestimation of carbon monoxide (CO) concentrations at extreme pollution events in the deep learning (DL)
3.1 CO concentrations predicted by DL model
It seems that the rapid decrease of surface CO concentrations over NCP 2019 is associated with an unexpected drop in CO emissions, which is not considered in the linear projection of emission inventory, and led to overestimated CO concentrations in the DL model
Summary
Accurate simulation and prediction of air pollutants are critical for making effective policies to improve air quality. Based on CTMs, data assimilation techniques integrate simulations and observations and can improve the modeled atmospheric compositions. Despite the advantages of the DL approaches, the lack of parameterization of physical and chemical processes implies the predicted atmospheric compositions may deviate from the realistic atmospheric state, in contrast to conventional data assimilation approaches that are constrained by modeled processes. We perform a comparative analysis between the DL model and a KF system in this work, to investigate the performances of the two approaches in predicting CO This comparison is helpful for understanding the advantages and disadvantages of the DL approach in respective to traditional data assimilation, which is critical for better applications of this novel technique in atmospheric environmental studies in the future.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
