the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Sensitivity of the tropical Atlantic to vertical mixing in two ocean models (ICON-O v2.6.6 and FESOM v2.5)
Abstract. Ocean General Circulation Models still have large upper-ocean biases e.g. in tropical sea surface temperature, possibly connected to the representation of vertical mixing. In earlier studies, the ocean vertical mixing parameterisation has usually been tuned for a specific site or only within a specific model. We present here a systematic comparison of the effects of changes in the vertical mixing scheme in two different global ocean models, ICON-O and FESOM, run at a horizontal resolution of 10 km in the tropical Atlantic. We test two commonly used vertical mixing schemes; the K-Profile Parameterisation (KPP) and the Turbulent Kinetic Energy (TKE) scheme. Additionally, we vary tuning parameters in both schemes, and test the addition of Langmuir turbulence in the TKE scheme. We show that the biases of mean sea surface temperature, subsurface temperature, subsurface currents and mixed layer depth differ more between the two models than between runs with different mixing scheme settings within each model. For ICON-O, there is a larger difference between TKE and KPP than for FESOM. In both models, varying the tuning parameters hardly affects the pattern and magnitude of the mean state biases. For the representation of smaller scale variability like the diurnal cycle or inertial waves, the choice of the mixing scheme can matter: the diurnally enhanced penetration of equatorial turbulence below the mixed layer is only simulated with TKE, not with KPP. However, tuning of the parameters within the mixing schemes does not lead to large improvements for these processes. We conclude that a substantial part of the upper ocean tropical Atlantic biases is not sensitive to details of the vertical mixing scheme.
- Preprint
(6359 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 25 Sep 2024)
-
CC1: 'referee comment', Gilles Reverdin, 13 Aug 2024
reply
The paper investigates the impact of turbulence parameterizations in ocean models, focusing on the equatorial Atlantic in 2015. Two ocean models that are also part of climate models are considered: ICON-O and FESOM. They both have intermediate high horizontal and vertical resolution (128 layers), but with very different horizontal grids and schemes. Both have a z* vertical grid (with SSH link, but not quite the same, if I correctly understood). However, they have rather comparable near-equatorial horizontal resolution. These are forced runs (ERA5 forcing terms, mostly) with use of bulk formula for the air-sea exchanges. Runs are typically done on two years: different turbulent schemes are tested (in particular In ICON-O) as well as the bulk formula formulation (to test the influence on the runs of the differences in flux formulation between the two). The test runs are two years long, the second year been considered, which should be enough for the near equatorial adjustment, but avoid the larger basin scale adjustments that will result from the different turbulence parameterizations. Notice also that most of the changes made should mostly modify the near surface mixing, and not so ‘directly’ the deeper one. The bulk formulas used both imply negative feedback towards the ERA5 atmosphere temperature in the tropical Atlantic. However, in regions where the model produce excess surface temperature (such as south-eastern equatorial Atlantic), either because of the other components of the heat budget (radiative…) or because of the turbulence scheme or model simulations, this would add a destabilizing term in the mixed layer, and thus moderate excess near surface stratification. In regions, where the models are too cold (probably too much upwelling or thermocline structure not well reproduced), this would contribute to some added stratifying term, and thus very reduced MLD. I am just stating that as there could be a link between SST biases and MLD bias structures related to the overall bulk formulation (which ever of the two is used). The main results are that MLD underestimation bias (and SST too low bias) is overall large (almost a factor of two in some areas and runs; most noticeable between 0 and 10°S in central and western Atlantic), and although sensitive to the turbulence parameterization, not to the point of changing main patterns (same for the SST bias structure). There is some dependency nonetheless on scheme which is explained, and systematic differences remaining between the two models. As pointed out by the authors, some of this bias might be due to overall thermocline structure and flow. There is then ad good discussion on high frequency variability in particular the diurnal warm layer (but also inertial waves). This is likely important, due to modulation and possible impact on momentum flux in the ocean (less so for heat and water flux, as it is more linear…). This made me wonder about the ‘in situ’ reference used for those. It is based on Argo data using (according to figure 1 caption) all Argo profiles in 2000 to 2022), which implies that a large part of the profiles were not starting above 7 m, and thus missing most of the DWL. There is also a question on when in the day the Argo profiles arrive at the sea surface, which is often not homogeneous across the tropical Atlantic or through the day. Are the authors sure that there is no systematic difference due to that the in-situ Argo data reference, and by how much (and could there be a spatial structure in it due to distribution of timing of Argo profiles through the day). At least, this is not consistent with what is presented from the model, which uses a surface reference for density (a density criterium is used). In some ways, it could have been preferable to use for the models a late-night value to compare with the Argo climatology (and for those always taking the reference near 7 m, or other fixed depth). Whereas the investigation of DWL and diurnal cycle require another analysis (as is done). For SST, it is HADISST which is used as a reference. It would be important to remind whether it is the daily average SST which is used for the comparison (or something else). There are also interesting results on near equatorial mixing and day time of maximum diffusivity, deep cycle turbulence, with suggestions of some of the TKE (or PKK) runs performing more satisfactorily than the others (for these investigations, other data sets are used, which seem appropriate for the investigation, as well as fr the off equatorial DWL) (well for deep-cycle turbulence it is less so, based on figure 13 and 14) Overall, would it be fair to say that somehow, we have two models with rather large systematic large-scale biases (as also seen in Figure 7) that would not change much in such a short term as 1 to 2 years of the tests, which mostly tackle the surface mixing (although some minimum values also impact the subsurface terms). Maybe that could be a reason with the differences in the overall results with what is found in Deppenmeier et al (2020) investigating ck dependency of bias in NEMO (larger ck leading to surface cooling and subsurface warming, and less SST bias). Alltogether it is a rather interesting study worth publishing. Minor comments. For TKE they use Pr=6.6Ri which I find large (I believe that it is 1 in NEMO, but have not checked). Why this choice? l. 372: ‘to a large extent on the surface velocity of the ocean’ (‘to a large extent’ may be a little too strong; more for energy than for heat/water). After, I understood the point made on the relative direction of wind and currents, and thus the difference between the equatorial and off-equatorial situations, but there to impact larger for energy/wind power than for heat/water (and it is not so clearly separated, according to a recent paper, Hans et al (2024)) The authors are likely aware of the Hans et al. paper (JGR Oceans in Press) that carefully evaluates from data the structure of the DWL and its diurnal jet along the equatorial Atlantic, and could complement what is discussed in the paper. Figure 2: I would write instead: “Annual mean mixed layer depth in 2015 (correct?) for the different simulations of FESOM and ICON-O relative to the Argo climatology (for 2000-2002?) presented on the left top panel. (or difference in annual MLD…, but not ‘between FESOM and ICON-O runs’) Captions of figure 7 and 8 incorrect. The panels show SST difference (except for the Argo one). Figure 7 caption: over which latitudinal band are the 2015 Argo profiles (the reference for the other panels) averaged. Does this averaging scale have an impact (or not) on the anomalies presented on the other panels for the different model runs. Altogether, some figure captions are not very detailed (and information has to be retrieved from the core of the paper to figure to what they correspond. Another example is figure 11, in which no special domain is specified for the model runs, nor where the observations were collected. Figure 10, I understand what is attempted, but I have a hard time looking at it, convincing myself on what is said in the paper. In this case, is it important to show all the panels. I can imagine many reasons that may not be that relevant for the overall conclusions, why the model runs don’t reproduce the special event found in the data. Figure 12: I understood afterwards the choice of days 120-138 of year 2014 (reading the paper). On the other hand, the dates are rather close to the beginning of the test simulations, and could be sensitive to it. The results are very different between the runs, for example the lower (and not daily?) modulation in I_KPP_01 to 03. I did not fully understand what is from it the lesson. Why is there this 5-day modulation in these three runs and not in the others; Would they have had some ‘instability’ waves, for example, that are not present in the other runs. And from those panels, how does one feel what is expected? (the last sentence of the caption is for that a bit vague, and not informative) Fig. 15: Is what is shown the results in ICON-O of using the alternative bulk formula (and with comparison to the observed MLD, as in Fig. 2). Or is it instead the difference of the two sets of runs which the caption would suggest. Figure 17: I assume that only the third panel from left with the alternate forcing bulk formulae in ICON-O. I would remove in the title ‘Effect of exchanging’… and be more specific on what are the runs presented…
Citation: https://doi.org/10.5194/egusphere-2024-2281-CC1
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
242 | 50 | 15 | 307 | 7 | 4 |
- HTML: 242
- PDF: 50
- XML: 15
- Total: 307
- BibTeX: 7
- EndNote: 4
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1