Spatial regression analysis on 32 years of total column ozone data

• Published in: Atmospheric Chemistry and Physics
• Main Authors: Knibbe, J.S., van der A, J.R., de Laat, A.T.J.
• Format: Article in Journal/Newspaper
• Language: English
• Published: 2014
• Subjects: Antarctic; Pacific; The Antarctic; Antarc*
• Online Access: https://research.vu.nl/en/publications/24cc26f8-566c-4caa-a93f-5bcf201247a5; https://doi.org/10.5194/acp-14-8461-2014; http://hdl.handle.net/1871.1/24cc26f8-566c-4caa-a93f-5bcf201247a5

Multiple-regression analyses have been performed on 32 years of total ozone column data that was spatially gridded with a 1 × 1.5° resolution. The total ozone data consist of the MSR (Multi Sensor Reanalysis; 1979-2008) and 2 years of assimilated SCIAMACHY (SCanning Imaging Absorption spectroMeter for Atmospheric CHartographY) ozone data (2009-2010). The two-dimensionality in this data set allows us to perform the regressions locally and investigate spatial patterns of regression coefficients and their explanatory power. Seasonal dependencies of ozone on regressors are included in the analysis. A new physically oriented model is developed to parameterize stratospheric ozone. Ozone variations on nonseasonal timescales are parameterized by explanatory variables describing the solar cycle, stratospheric aerosols, the quasi-biennial oscillation (QBO), El Niño-Southern Oscillation (ENSO) and stratospheric alternative halogens which are parameterized by the effective equivalent stratospheric chlorine (EESC). For several explanatory variables, seasonally adjusted versions of these explanatory variables are constructed to account for the difference in their effect on ozone throughout the year. To account for seasonal variation in ozone, explanatory variables describing the polar vortex, geopotential height, potential vorticity and average day length are included. Results of this regression model are compared to that of a similar analysis based on a more commonly applied statistically oriented model. The physically oriented model provides spatial patterns in the regression results for each explanatory variable. The EESC has a significant depleting effect on ozone at mid-and high latitudes, the solar cycle affects ozone positively mostly in the Southern Hemisphere, stratospheric aerosols affect ozone negatively at high northern latitudes, the effect of QBO is positive and negative in the tropics and mid-to high latitudes, respectively, and ENSO affects ozone negatively between 30° N and 30° S, particularly over the Pacific. The contribution of explanatory variables describing seasonal ozone variation is generally large at mid-to high latitudes. We observe ozone increases with potential vorticity and day length and ozone decreases with geopotential height and variable ozone effects due to the polar vortex in regions to the north and south of the polar vortices. Recovery of ozone is identified globally. However, recovery rates and uncertainties strongly depend on choices that can be made in defining the explanatory variables. The application of several trend models, each with their own pros and cons, yields a large range of recovery rate estimates. Overall these results suggest that care has to be taken in determining ozone recovery rates, in particular for the Antarctic ozone hole. © 2014 Author(s).