Adaptive Aerosol Optical Depth Forecasting Model Using GNSS Observation

As one of the important factors in atmospheric physical and chemical processes, aerosol optical depth (AOD) has an important impact on regional and global climate. Therefore, monitoring and predicting the temporal and spatial changes of AOD is of considerable significance. Existing methods mainly use a large number of meteorological parameters and ground observations to forecast AOD. However, modeling data are numerous and difficult to obtain practically. In this study, an adaptive AOD forecasting (AAF) model is proposed using the zenith total delay (ZTD) derived from global navigation satellite system (GNSS). This model only uses the ZTD as the external input parameter and considers the time autocorrelation of AOD for the previous epoch. In addition, AAF can adaptively adjust the model coefficients and has high accuracy. The AOD data derived from the Second Modern-era Retrospective Analysis for Research and Applications, Version 2 (MERRA-2) and Aerosol Robotic Network in the Beijing–Tianjin–Hebei (BTH, $113^{\circ } 27^{\prime }$ E– $119^{\circ } 50^{\prime }$ E, $36^{\circ } 05^{\prime }$ N– $42^{\circ } 40^{\prime }$ N) region over the period of 2015–2017 are used to perform the experiment. In addition, ZTD data of 16 GNSS stations in BTH region from the Crustal Movement Observation Network of China are selected to establish the AAF model. Experimental result reveals good performance of the proposed AAF model for internal and external validations. The difference in root mean square (rms), mean absolute error, and Bias of AOD between the AAF model and MERRA-2 are 0.11, 0.08, and 0.03, respectively. Compared with the existing AOD forecast models, the proposed AAF model is superior in terms of time resolution, rms, and correlation.