| 08 Sep 2026
Parameterized orographic gravity wave forcing in the valve layer affects boreal stratospheric polar vortex in CMIP6
Abstract. The stratospheric polar vortex critically influences wintertime stratospheric circulation, as well as troposphere-stratosphere coupling and trace gas distribution. Reproducing correct statistics of stability and strength of the polar vortex is an important challenge for Earth system models as its variability affects regional climate predictability. The latest generation of the models are showing pronounced biases and intermodel spread in this regard. Here, we show how parameterized orographic gravity wave drag (OGWD) in climate models modulates planetary wave propagation, impacting SPV stability and morphology. We analyze simulations from 14 models produced within the Coupled Model Intercomparison Project Phase 6 (CMIP6). The orographic gravity wave drag varies between the models, depending on a parameterization scheme and the tuning of free parameters. We investigate how the intermodel differences in the parameterized gravity wave drag in the upper troposphere lower stratosphere relate to resolved wave activity and SPV characteristics. We find that stronger OGWD suppresses upward planetary wave propagation, leading to reduced wave breaking at the SPV edge and a more stable, broader vortex. By documenting the role of the orographic gravity wave drag for the boreal stratospheric polar vortex in the models, we highlight the potential of improvements or careful tuning of this parameterization for alleviating the climate model biases related to polar vortices.
The manuscript ”Parameterized orographic gravity wave forcing in the valve layer affects boreal stratospheric polar vortex in CMIP6” by Hàjkovà et al. analyses the relationships between parameterized OGWD and stratospheric dynamics in 14 CMIP6 models. The authors demonstrate that a larger value of OGWD in the subtropical lower stratosphere (the valve layer) are associated with reduced upward and equatorward propagation of the planetary waves, increased wind speed at the equatorward flank of the polar night jet and reduced frequency of disturbed polar vortices, both displaced and elongated. The authors explain it by the compensation mechanism – changes in the parameterized OGWD typically lead to opposing changes in resolved wave drag.
The paper is written clearly; however, I am not sure whether the results are particularly novel. The effect of the compensation mechanism has previously been demonstrated for CMIP6 models for example by Hàjkovà and Šàcha (2024). Also, I am concerned that the causality of the relationship between OGWD and stratospheric winds is not sufficiently demonstrated. Therefore, I cannot recommend the manuscript for publication in its present form. Below are my specific comments.
Major comments:
1. It is not clear to me what is the novelty and significance of this manuscript. There is a discussion of how OGWD is important for SSW but the stated goal of the manuscript is to investigate “whether the intermodel differences in the boreal stratospheric SPV climatology and stability in CMIP6 can also be traced back to the differences in climatological parameterized OGWD.” I am not convinced it is much different from Hàjkovà and Šàcha (2024) who “has shown that the climatological intermodel differences of stratospheric dynamics in CMIP6 models are strongly correlated with the differences in the climatological parameterized OGWD in the lower stratosphere, especially in the valve layer.” I suggest the authors make it clearer what the results of this manuscript add to the existing knowledge and why it is important.
2. I find it somewhat disappointing that the authors do not try to connect their results to the properties of the parameterisation schemes used in the models. Ignoring this makes it difficult to establish causality - how certain is that the OGWD in the valve layers affects the properties of the zonal winds and resolved waves, and not the other way round? Gravity waves break in the lower stratosphere because of the strong wind gradients there. Could it be so that factors other than OGWD control wind gradients in the valve layer, and these wind gradients in turn affect gravity wave breaking and OGWD? I feel that you need to show that the properties of the GW parameterization schemes control the amount of OGWD in the valve regions. Without this the results are inconclusive. I suggest the authors link their results to the type of the parameterisation scheme used, and if possible to their properties, such as amount of gravity wave activity produced at the surface, criteria for wave breaking etc.
Other comments:
Figure 1: Figure 1 shows SSW statistics across the models but it is not discussed in the rest of the manuscript – how is SSW statistics related to OGWD? I think Figure 1 can be removed since its value for the manuscript is not clear.
L29: “the morphology of the edge of the SPV and the structure of potential vorticity can be used to describe the stability of the SPV”
I would say that the gradient of PV across the edge of the polar vortex can be used to quantify vortex stability. The current text does not sound clear to me, perhaps you need to rewrite it.
L31: I would add Charlton and Polvani (2007) to the reference list: Charlton, A., and L. Polvani, 2007: A new look at stratospheric sudden warmings. Part I: Climatology and modeling benchmarks. J. Climate, 20, 449–469.
L46: “the climatological intermodel differences of stratospheric dynamics in CMIP6 models are strongly correlated with the differences in the climatological parameterized OGWD in the lower stratosphere”
I think this needs to be explained better – what is correlated with what?
L56: What is the benefit of using AMIP experiments? Don’t you introduce unphysical biases due to the lack of atmosphere-ocean coupling?
L61: I do not think the sentence should start with “However,…”
L61 “the ensemble models” – do you mean “across ensemble members”?
L70: I suggest referring to DynVarMIP (Gerber and Manzini 2016), which made the dynamical outputs from CMIP6 possible:
Gerber, E. P. and Manzini, E.: The Dynamics and Variability Model Intercomparison Project (DynVarMIP) for CMIP6: assessing the stratosphere–troposphere system, Geosci. Model Dev., 9, 3413–3425, https://doi.org/10.5194/gmd-9-3413-2016, 2016.
L86-87: Do you calculate the vertical and meridional components of the EP-flux using data on pressure levels? How accurate these estimates are? You calculate the vertical gradient of potential temperature, which is probably not very accurate given the sparse vertical resolution of pressure levels. Are not the EP-flux components available from the CMIP6 database as a part of the DynVarMIP output?
L100: I am missing details of PV calculations. Do you calculate relative vorticity on isentropic levels? Are winds available on isentropic levels or do you interpolate winds from pressure levels onto isentropic levels? Do you calculate vertical gradient of potential temperature between pressure levels or between model levels? If it is between pressure levels then how accurate these vertical gradient estimates are? Is model data interpolated on the same horizontal grid?
L117: I suggest replacing “relationship” with “correlation” for clarity, because this is what is shown in Figure 3.
L117: To motivate the selection of the “valve layer” I suggest a latitude-pressure figure with multimodel mean OGWD. I assume the “valve layer” would show an enhanced values of OGWD.
L120: What do you mean by “absolute correlation” in this context?
L130: “In the CMIP6 models that relationship was documented by Hajkova and Šacha (2024) to govern the intermodel differences in the boreal stratospheric dynamics.”
This sentence should be reworded. It is not clear.
Figure 5a: The vertical component of the EP-flux in pressure coordinates, as given by Eq. 1, should be negative, because the potential temperature gradient with respect to pressure is negative. Is it negative in Figure 5a?
L212: “However, it has to be pointed out that the results presented here do not provide direct conclusions regarding SSWs and the intermodel differences in them. Naturally, we have tried to extend the analysis also towards this direction, but the results were not as clear as the ones presented in this manuscript.”
This text is not very informative. Can you explain what exactly is not clear?