A numerical weather prediction approach to assessing the stratospheric impact on tropospheric predictability

Recent observational studies have suggested that stratospheric variability may have a significant impact on the predictability of tropospheric modes of variability, such as the North Atlantic Oscillation, on timescales of weeks to months. It has been argued that such improved skills can be realized by including a well-resolved stratosphere in forecast models, but actual comparison of forecasts with and without stratospheric information have not, to our knowledge, been performed to date.

We here attempt to quantify the impact of the stratosphere on tropospheric predictability using a comprehensive general circulation model of the troposphere-stratosphere system being developed at NCAR, namely the Whole Atmosphere Community Climate Model (WACCM). Although WACCM is primarily designed to be a climate model, we use it in forecast mode and in a "perfect model" context, i.e. we consider a long control run of WACCM to be the "truth", and try to predict the evolution of the control run using "perturbed" runs with different initial stratospheric conditions and model configurations.

One class of perturbed runs involves degrading the initial condition in the stratosphere to climatology, in order to determine if and how the quality of the tropospheric forecasts is also degraded. In these runs, the full stratosphere is still part of the model, and thus the dynamical coupling between the two is still present (e.g. waves emanating from the troposphere are able to propagate into the climatological stratosphere, and possibly back to the troposphere at a later time to affect the forecasts). A second class of perturbed runs involves eliminating the stratosphere altogether from the model by lowering the upper boundary and thus limiting the vertical domain to the troposphere alone. In these runs the actual dynamical coupling between the stratosphere and troposphere is completely altered. We will report on a suite of ensemble forecasts using both classes of perturbations.