The influence of stratospheric temperature changes on ozone trends: Analysis of OMI ozone products and improvements for the differential optical absorption spectroscopy (DOAS) technique that is applied to OMI satellite measurements.

Abstract

The objectives of this study are 1. to validate the OMDOAO3 fitted effective temperature; 2. to estimate the temperature sensitivity of OMI ozone products and 3. improve the DOAS concept such that an ozone trend can accurately be determined with DOAS obtained vertical ozone column amounts. It has been determined that an ozone retrieval that is able to capture ozone trends with an accuracy of 1%, needs to have a temperature sensitivity of approximately 0.01%/K or less. For the Brewer direct sun observations, the temperature sensitivity is estimated with the latest ozone absorption cross sections from Serdyuchenko as 0.014+/-0.003%/K. Hence that the Brewer direct sun observations are not sensitive to changes in effective ozone temperature, which is also observed with 1998-1999 Toronto measurements in a study of [1, Kerr, J. B., 2002] , and can therefore be used to validate OMI ozone products. The ozone profile from OMO3PR is cross-validated with other ozone profiles in a study of [2, M. Kroon et al., 2011] and therefore an ozone effective temperature can be accurately determined with an ozone profile from OMO3PR and a temperature profile from medium-range weather forecasting model (ECMWF). This effective temperature can be used as a reference effective ozone temperature, to validate the OMDOAO3 fitted effective temperature, and to estimate temperature sensitivities of OMI ozone products. For OMDOAO3 the effective temperature is fitted from the spectrum itself. It is found that the fitted effective temperature from OMDOAO3 has an offset of -5.82+/-0.04 C, which is consistent for different seasons and regions. This proves the concept that it is possible to retrieve an ozone effective temperature from the spectrum itself with the differential optical absorption spectroscopy (DOAS) technique, which in principle should give an ozone column amount independent of temperature. For OMTO3, the other OMI ozone column amount product, temperature profiles are used inconsistently with the ozone profiles. The TOMS v8 climatology for temperature profiles depends on month and latitude, and therefore errors in effective ozone temperature depending on latitude or season are not observed. However, other variations in effective ozone temperature are not captured by the climatology, which are: stratospheric climate change, volcanic eruptions and longitudinal temperature variations. From a cross-validation of OMI ozone products, the temperature sensitivity can be estimated for OMI ozone products, which varies between 0.06 and 0.13 %/K in absolute magnitude. This is higher than the threshold value of 0.01%/K for an ozone retrieval that is insensitive to temperature variations, and it can be concluded that errors in ozone trends can be expected on the order of 10% for OMI ozone products. The temperature sensitivity can also be determined with the help of ground-based measurements. When the temperature sensitivity for OMTO3 is determined, by comparing the difference in column amount to Brewer direct sun observations, against effective temperature difference, a temperature sensitivity of 0.255%/K is found, which is probably caused by longitudinal temperature variations that are not captured by the OMTO3 algorithm. For all OMI ozone products a temperature sensitivity is found, when the ozone column amounts are compared to the Brewer direct sun observations as function of temperature. This may be caused by the choice of ozone absorption cross sections and how its temperature dependency is implemented in the algorithm. For OMO3PR a temperature sensitivity of 0.0720 %/K is found, and for OMTO3 a temperature sensitivity of 0.112 %/K. For OMDOAO3 a temperature sensitivity of 0.0275 %/K is found, which shows that although an effective temperature may be fitted from spectrum, the ozone column amount itself can still be sensitive to temperature variations. 1 For the current OMDOAO3 model function and fit window, the fitted effective temperature, compared to the modeled slant effective temperature, has an offset of −5.8+/-1.7 C in the simulations. This is in good agreement with the offset that has been found with observations for the OMDOAO3 fitted effective temperature of -5.82 +/- 0.04 C. Hence that the found offset in effective ozone temperature can be explained by the DOAS fitting method itself. Considerations for improving the OMDOAO3 algorithm are done, where the main improvements are 1. the use of all temperature expansion coefficients of the second-degree polynomial of the ozone absorption cross sections instead of linearizing them - and 2. the use of a wavelength-dependent slant column amount. Simulations show that the first improvement can reduce the temperature sensitivity of the algorithm by a factor of ten, which makes the DOAS concept a reliable source for ozone trend determination. The second improvement, the use of a wavelength dependent slant column amount, enables the use of 17.5 nm wide fit window, about four times as wide as the current OMDOAO3 fit window, which can enhance noise reduction, resulting in an ozone product with a better overall resolution.