Study On Quality Control And Assimilation Of Satellite Ozone Data And The Application In The Simulation Of Tropical Cyclone

Since ozone concentration varies at different altitudes and latitudes due to inhomogeneous transport of atmospheric flow, total column ozone is a passive tracer on synoptic scale. Therefore, ozone is highly correlated with many atmospheric variables. Based on the correlation between total column ozone and potential vorticity (PV), this study has made an attempt to develop the quality control (QC) scheme of Atmospheric Infrared Sounder (AIRS) ozone data and Fengyun-3A (FY-3A) Total Ozone Unit (TOU) ozone data in the first step. Following this step, the four-dimensional variational (4D-Var) assimilation scheme of AIRS ozone data is adopted and implemented in the mesoscale non-hydrostatic model MM5 to improve the numerical prediction of Tropical Cyclones (TCs). The results are summarized as follows:(1) It was found that almost all of the ozone data within TCs were flagged to be of bad quality by the AIRS original QC scheme. This determination was based on the ratio of total precipitable water (TPW) error divided by TPW value, where TPW was an AIRS retrieval product. It was found that the difficulty in finding total column ozone data that could pass AIRS QC was related to the low TPW employed in the AIRS QC algorithm. In this paper, a new two-step QC scheme for AIRS total column ozone is developed. A new ratio is defined which replaces the AIRS TPW with the zonal mean TPW retrieved from the Advanced Microwave Sounding Unit (AMSU). The first QC step is to remove outliers when the new ratio exceeds 33%. Linear regression models between total column ozone and mean potential vorticity (MPV) are subsequently developed with daily updates, which are required for future applications of the proposed total ozone QC algorithm to vortex initialization:and assimilation of AIRS data. In the second QC step, observations that significantly deviate from the models are further removed using a biweighting algorithm. Numerical results for two typhoon cases and two hurricane cases show that a large amount of good quality AIRS total ozone data is kept within TCs after implementing the proposed QC algorithm.(2) To apply the TOU ozone data in numerical weather prediction, a QC scheme is developed, viewed from the assimilation point. In the first step, a daily updated linear regression model, which links the total column ozone to the MPV, is established. Following this step, the biweight algorithm is applied to remove the outliers. Numerical results implementing the proposed QC scheme in typhoon Tembin (2012) and Isaac (2012) reflect daily variations of correlation between total ozone and MPV. The total percentage of outliers identified by this scheme is highly stable with the change of time, and the main information of ozone data is maintained whereas the standard deviation is reduced significantly. In addition, the ozone data after QC are more consistent with the statistical fitting variable. The distribution of the observed-minus-fitting Probability Density Function becomes nearly Gaussian, which is conducive to data assimilation.(3) Based on the correlation between total column ozone and PV, the observation operator of each level is established for the 4DVAR assimilation of the AIRS ozone observations. Results from the numerical experiments using the proposed assimilation scheme for Hurricane Earl show that the ozone data assimilation affects the PV values at high level first and then influences those at middle and low levels. Through the so-called asymmetric penetration flows associated with the upper-level PV anomaly, the modification of the PV field after the assimilation of AIRS ozone data improves the track and intensity prediction of Hurricane Earl. With the AIRS ozone data being assimilated, the warm core of Hurricane Earl is intensified, resulting in the improvement of other fields near the hurricane center, and the track prediction is improved mainly due to the adjustment of the steering flows in the assimilation experiment.(4) This study develops the bogus and ozone data assimilation (BODA) scheme for improving hurricane prediction. First, the observation operator at some model levels with the highest correlation coefficients is established to assimilate AIRS ozone data based on the correlation between total column ozone and PV ranging from 400 hPa to 50 hPa level. Second, AIRS ozone data act as an augmentation to a bogus data assimilation (BDA) procedure using a 4D-Var data assimilation system. Case-studies of several hurricanes are performed to demonstrate the effectiveness of BODA scheme. The statistical result indicates that assimilating AIRS ozone data at 4,5, or 6 model levels can produce a significant improvement in hurricane track and intensity prediction, with reasonable computation time for the hurricane initialization. It is found that the new scheme developed in this study generates significant adjustments in the initial conditions (ICs) from the lower levels to the upper levels, compared with the BDA scheme. With the BODA scheme, hurricane development is found to be much more sensitive to the number of ozone data assimilation levels. In particular, the experiment with the assimilation of AIRS ozone data at proper number of model levels shows great capabilities in reproducing the intensity and intensity changes of Hurricane Earl, as well as improve the track prediction. These results suggest that AIRS ozone data convey valuable meteorological information in the upper troposphere, which can be assimilated into a numerical model to improve hurricane initialization when the low-level bogus data are included.