Impacts on SH tropospheric circulation predictability of incrementing stratosphere vertical resolution in climate models

MARISOL OSMAN; CAROLINA VERA

Queenstown

Conferencia; SPARC GA 2014; 2014

WCRP-SPARC

The study documents the predictability of seasonal means of 200-hPa and 850-hPa geopotential heights in the Southern Hemisphere (SH) during winter and summer. Ensembles of lead-1-month climate predictions outputs for DJF and JJA available from the WCRP/CHFP-SHFP Project are used. In particular, 4 different climate models with ?high-top? versions, which resolve the stratosphere, and ?low-top? versions, which do not, are considered. Predictability is defined as the ratio between the interannual standard deviation of the ensemble mean (S) and the ensemble spread (N). Predictability changes associated with changes in model vertical resolutions are described by computing the ratio between high-top model predictability and low-top model predictability. The predictability analysis performed for each of the models considered, shows that for all of them, signal is larger over the Bellingshausen-Amundsen seas for all seasons and levels considered. Also, some models present at around 45ºS significant signal values, but its zonal location is model dependent. In DJF, predictability at high levels is also large at tropical and subtropical latitudes in the South Pacific, likely related with the atmospheric response to ENSO. Nevertheless, that signal is not evident in JJA. At low levels, DJF signal is also large over northern Australia and, for some models, over the Maritime Continent too. On the other hand, noise is large at middle and high latitudes and, in general, is larger (smaller) in DJF than in JJA at high (low) levels. Predictability changes between models resolving the stratosphere and those who do not, are model dependent for all seasons and levels considered. CMAM model presents in general alternating regions of increased and reduced predictability, concentrated at high latitudes of the South Pacific, and mainly associated with signal changes. On the other hand, ARPEGE model shows in general no improvement of SH circulation predictability with better resolution of stratosphere. Exceptions are the southern tip of South America at low levels and the vicinity of the Antarctic Peninsula at both levels, where predictability improvement is found for the ARPEGE high-top version. In addition, IFS high-top model clearly improves predictability at high levels in JJA, mainly associated with noise reduction. Although, at low levels and particularly in DJF, predictability provided by this model largely decreases at the tropics. Finally, predictability improvements provided by the HadGem high-top model, are larger in JJA than in DJF. Such changes, which are more evident at high levels, are mainly associated with noise reductions at tropical regions and signal increases at middle and high latitudes.