Interannual-to-decadal variability of the stratosphere during the 20th century: ensemble simulations with a chemistry-climate model

Interannual-to-decadal variability in stratospheric ozone and climate have a number of common sources, such as variations in solar irradiance, stratospheric aerosol loading due to volcanic eruptions, El Nino Southern Oscillation variability and the quasi-biennial oscillation (QBO). Currently available data records as well as model simulations addressing stratospheric chemical climate variability mostly cover only the past few decades, which is often insufficient to address natural interannual-to-decadal variability. Here we make use of recently reconstructed and re-evaluated data products to force and validate transient ensemble model simulations (nine members) across the twentieth century computed by means of the chemistry-climate model SOCOL (SOlar Climate Ozone Links). The forcings include sea surface temperatures, sea ice, solar irradiance, stratospheric aerosols, QBO, changes in land properties, greenhouse gases, ozone depleting substances, and emissions of carbon monoxides, and nitrogen oxides. The transient simulations are in good agreement with observations, reconstructions and reanalyses and allow quantification of interannual-to-decadal variability during the 20th century. All ensemble members are able to capture the low-frequency variability in tropical and mid-latitude total ozone as well as in the strength of the subtropical jet, suggesting a realistic response to external forcings in this area. The region of the northern polar vortex exhibits a large internal variability that is found in the frequency, seasonality, and strength of major warmings as well as in the strength of the modeled polar vortex. Results from process-oriented analysis, such as correlation between the vertical Eliassen Palm flux (EP flux) component and polar variables as well as stratospheric ozone trends, are of comparable magnitude to those observed and are consistent in all analysed ensemble members. Yet, trend estimates of the vertical EP flux component vary greatly among ensemble members precluding any robust conclusions. This suggests that internal variability in models must be accounted for in order to quantify the atmospheric model response in wave energy upon external forcings.