Tracking sudden stratospheric warming (SSW) onset with the adjoint

Daniel Holdaway and Lawrence Coy

The winter stratosphere (~18-55 km in altitude) is characterized by a strong westerly jet (~50-100 m/s) circulating about the cold polar night.  The strength and location of this jet is highly variable, especially in the Northern Hemisphere (NH) where land/ocean heating contrasts force tropospheric jet stream disturbances (our weather) that can propagate vertically into the stratosphere as large, planetary-scale waves (PW).  Every 2-4 years in the NH, the temperature in the polar night stratosphere increases rapidly (~40° K in ~5 Days) and the westerly stratospheric jet breaks down as the PWs increase in amplitude in response to elevated tropospheric forcing.  These events are known as major Sudden Stratospheric Warmings (SSW).  Only one major SSW event (September 2002) has been observed in the ocean dominated SH.  While tropospheric jet variations are generally understood to force the SSW the exact evolution and structure of the tropospheric forcing varies for different SSW events and is still an open question.

In this work we have developed a framework, using the global GEOS-5 adjoint model, to identify the tropospheric weather systems that directly lead to a SSW. The mathematical framework of the adjoint model is capable of identifying the origin of observed features, such as the warm polar night temperatures associated with a SSW event.  This provides an efficient mechanism for investigating the tropospheric forcing (the origin) of rare events such as the September 2002 SH case, as well as identifying any general tropospheric weather patterns that occur ahead of all SSWs.

In the example presented in the slide the SH major SSW of September 2002 is shown. The slide shows two forecasts for the same period. The top row shows a standard GEOS-5 global forecast, initialized from the MERRA-2 reanalysis. The bottom row shows a new forecast where the initial conditions of the model are optimally perturbed as determined by the GEOS-5 adjoint model.  By incorporating the optimal perturbation provided by the GEOS-5 adjoint model the major SSW can effectively be turned off.  These figures show the end result in the stratosphere but by tracking how the perturbation evolves from the perturbed initial state we can examine the origin and structure of the tropospheric PWs that led to the SSW. From these results we can further our understanding of the conditions that precipitate SSWs and determine some of the unique features of the 2002 SH winter that resulted in the only known SH major SSW.

An outline of the adjoint model methodology:

  • Initialize the adjoint model with the mean temperature in the 2D box shown in the middle panels, i.e. the time and location at 10hPa where the warming is first identified.
  • Integrate the adjoint model backwards to answer the question: "what is the temperature in this box sensitive to 3 days before and in the troposphere only?"
  • Use Lagrange multipliers to convert the adjoint sensitivity into an optimal perturbation in the troposphere.
  • Run a new forecast with the initial fields perturbed by these adjoint derived fields.
  • Compare the original and perturbed forecast to see which mechanism is damped to prevent the SSW from happening. For example look at the difference between the running mean of standing wave fluxes or height anomalies.

« GMAO Science Snapshots