Sensitivity of stratospheric pathways of Arctic-midlatitude linkages to the modification of the gravity wave drag parameterization in ICON model using deep learning

The global warming has been observed to be more severe in the Arctic compared to the rest of the world. This enhanced warming i.e. Arctic Amplification is not just the result of local feedback processes in the Arctic. The stratospheric pathways of Arctic-midlatitude linkages and large-scale dynamical processes can contribute to the Arctic Amplification. The polar stratospheric dynamics crucially depends on the atmospheric waves at all scales. The winter polar vortex can be disturbed by gravity waves in the middle atmosphere. To investigate the sensitivity of the polar vortex dynamics, large-scale dynamical processes, and the stratospheric pathways of the Arctic-midlatitude linkages to the modification of gravity wave drag, we conduct sensitivity experiments using the global atmospheric model ICON-NWP (ICOsahedral Nonhydrostatic Model for Numerical Weather Prediction). These sensitivity experiments are performed by imposing a repeated annual cycle of the year 1985 for sea surface temperatures and sea ice as lower boundary conditions and for greenhouse gas concentrations as external forcing. This year is selected as both El-Nino Southern Oscillation and Pacific decadal oscillation were in their neutral phase and no explosive volcanic eruption has occurred. Hence, lower boundary and external forcing conditions in this year can serve as a useful proxy for the multi-year mean condition and an estimate of its internal variability. We performed simulations where in the control simulation the sub-grid parameterization scheme for both orographic and non-orographic gravity wave drags are switched on. The other two experiments are identical to the control simulation except that either orographic or non-orographic gravity wave drags are switched off.

    Recently, deep learning has extraordinarily progressed our ability to recognize complex patterns in big datasets. Deep neural networks have shown great capabilities to capture the dynamical process of the atmosphere. Applying deep learning algorithms on experiments’ results, the impact of gravity wave drag modifications on large-scale mechanisms of the Arctic Amplification will be analyzed. Special emphasis will be put on the effects of gravity wave drag modifications on the polar vortex dynamics.