Investigation of trends in tropical tropospheric ozone columns derived from CCD satellite data

Kurzfassung

With the convective cloud differential algorithm (CCD) the tropospheric ozone column (below 10 km) is calculated as the difference between the total ozone column for cloud free observations and the stratospheric column given by the vertical column above high reaching convective clouds. In this study our CCD algorithm is applied to the total ozone column product from GOME, SCIAMACHY, OMI and GOME-2. The resulting timeseries includes 20 years of tropospheric ozone data from 1995 to 2015. Here the same retrieval was applied to all instruments both for the total ozone column (GODFIT) as for the tropospheric column (CCD). This results in a very consitent data set. For the trend analysis individual ozone sonde stations as well as certain latitude bands were considered. Together with the linear term we fitted an El Niño index, a QBO index and a combination of several sinus and cosinus functions to simulate the seasonal variations as well as the phase shift. Despite the fact that the same algorithms were used, the observed trends still depend on the satellite instrument. While GOME-2 tropospheric ozone data showed a trend of 0.22 DU/yr in the 10°S to 0° latitude band between 2007 and 2014, the data from OMI increased by 0.037 DU/yr in the same latitude band and period. Assuming that the atmospheric trends are the same for the two overpass times of the instruments, we looked for instrumental effects as one possible explanation for the different trends. Therefore the data were compared to the SHADOZ ozone soundings. The partial column below 10 km was subtracted from the satellite insturments' data. The difference to the soundings showed temporal drifts for all instruments varying between -1.08 DU/yr and 0.06 DU/yr. The different trends in both the tropospheric satellite data and the validation indicates the need of further harmonized CCD data before analysing long-term trends.