Variability and trends in dynamical forcing of tropical lower stratospheric temperatures

The contribution of dynamical forcing to variations and trends in tropical lower stratospheric 70 hPa temperature for the period 1980-2011 is estimated based on ERA-Interim and Modern-Era Retrospective Analysis for Research and Applications (MERRA) reanalysis data. The dynamical forcing is estimated from the tropical mean residual upwelling calculated with the momentum balance equation, and with a simple proxy based on eddy heat fluxes averaged between 25° and 75° in both hemispheres. The thermodynamic energy equation with Newtonian cooling is used to relate the dynamical forcing to temperature. The deseasonalised, monthly mean time series of all four calculations are highly correlated (~ 0.85) with temperature for the period 1995-2011 when variations in radiatively active tracers are small. All four calculations provide additional support to previously noted prominent aspects of the temperature evolution 1980–2011: an anomalously strong dynamical cooling (~ −1 to −2 K) following the Pinatubo eruption that partially offsets the warming from enhanced aerosol, and a few years of enhanced dynamical cooling (~ −0.4 K) after October 2000 that contributes to the prominent drop in water entering the stratosphere at that time. The time series of dynamically forced temperature calculated with the same method are more highly correlated and have more similar trends than those from the same reanalysis but with different methods. For 1980–2011 (without volcanic periods), the eddy heat flux calculations give a dynamical cooling of ~ −0.1 to ~ −0.25 K decade⁻¹ (magnitude sensitive to latitude belt considered and reanalysis), largely due to increasing high latitude eddy heat flux trends in September and December-January. The eddy heat flux trends also explain the seasonality of temperature trends very well, with maximum cooling in January-February. Trends derived from momentum balance calculations show near-zero annual mean dynamical cooling, with weaker seasonal trends especially in December-January. These contradictory results arising from uncertainties in data and methods are discussed and put in context to previous analyses.