Use of Multiple Satellite Total Ozone Observations within and Around Tropical Cyclones

This study explores whether or not tropical cyclone (TC) structure information may be retrieved from satellite total ozone observations and how to link total ozone with analysis fields for potential application to TC vortex initialization schemes. Satellite total ozone retrievals from the Atmospheric Infrared Sounder (AIRS), the second Global Ozone Monitoring Experiment (GOME-2), the Ozone Monitoring Instrument (OMI), and the Ozone Mapping and Profiler Suite (OMPS) are used in this study. The study is divided into two portions: updating the quality control (QC) scheme for the AIRS total ozone data and examining applications of total ozone data in TC analyses. A modified QC scheme for AIRS total ozone is proposed to identify erroneous data while avoiding removal of potential useful data, as well as to keep the data consistent with a numerical prediction model. The modified QC scheme produces smaller bias and standard deviation of total ozone relative to the original AIRS QC scheme with less data being removed. Since ultraviolet (UV) retrieved total ozone generally suffers less cloud contamination, GOME-2 total ozone is used to examine possible TC structures captured by the data. All the TCs in the 2010-2012 Atlantic Hurricane seasons are used. Detailed comparisons of total ozone from GOME-2 and geopotential heights on isentropic surfaces from the European Centre for Medium-Range Weather Forecasts (ECMWF) Interim Reanalysis are carried out for the representative case of Hurricane Earl (2010). It is shown that GOME-2 total ozone can capture the characteristics of the storm center including the eye region and spiral bands. Changes in total ozone can also reveal the variations of the storm intensity. Alternatively, locally elevated total ozone in the hurricane eye region can be used to identify the upper tropospheric storm center, depending on the stage and the radius of the maximum winds of the storm. Further, the distribution of total ozone shows that the isentropic geopotential heights may have misplaced the vortex center. Total ozone and the 'aligned' geopotential heights at isentropic levels tend to form a linear relationship. Total ozone and geopotential heights correlate better in isentropic level range of 330K to 345K than at other isentropic levels, indicating better inference of the geopotential heights from total ozone at these isentropic levels. Stronger storms tend to have larger correlations of total ozone and geopotential heights at nearly all isentropic levels examined in this study. Inter-satellite calibration is shown to be necessary for better use of ozone observations in TC applications. A general inter-satellite total ozone calibration is carried out during August of 2012 with total ozone data from the four instruments. Total ozone from different instruments shows a dependence on cloud fraction, solar zenith angle (SZA), geo-location, and possibly aerosols. Overall, GOME-2, OMI and OMPS ozone observations agree well globally for both clear-sky and cloudy conditions, whereas AIRS shows large relative differences in both the southern Polar Regions and in the 30S-30N tropical regions. The large relative errors in the southern Polar Regions are associated with large AIRS SZAs, while the large relative errors in the 30S-30N zone may be caused by aerosol contamination and high cirrus clouds. Latitudinally varying coefficients are derived to remove large, zonally varying biases. Hurricane Michael (2012) is chosen to illustrate the resulting impacts of the inter-satellite total ozone calibrations on an individual TC's structure information. It is shown that the transformed observations are more comparable spatially and radially than the original ones. The impacts of the calibrated total ozone on the regression models linking isentropic geopotential heights and total ozone formed from all the TCs are examined. The calibration improves the performance of AIRS regression models at nearly all isentropic levels by increasing the explained variance and decreasing the root-mean-square-error (RMSE), while it does not change much with the regression models for OMI and GOME-2. Overall, stronger storms tend to have larger inference of geopotential heights at all the isentropic levels from these total ozone observations before and after the inter-satellite calibration. The inter-satellite total ozone calibration improves the comparability among different regression models. The resulting more consistent regression models based on TC's intensities can provide more consistent initial analysis fields from different total ozone observations for potential hurricane vortex initialization application.