Improve topography at 1deg and 025deg?

[aekiss]

At some point should we consider creating new topography for 1 and 0.25deg which is more consistent with the 0.1 deg topo?

The minimum depth is probably too large at 1deg and 025deg, now that we have finer surface resolution with KDS50: 45.11m (10 levels) in ACCESS-OM2, 40.36m (9 levels) in ACCESS-OM2-025, and 10.43m (7 levels) in ACCESS-OM2-01

The land masks are inconsistent, particularly near the tripoles:

There are also gaps in the depth distribution at 1 and 0.25 deg: https://github.com/COSIMA/access-om2/issues/141 and other problems at 1deg: https://github.com/mom-ocean/MOM5/issues/172 and non-advective points at 0.25 deg: https://github.com/COSIMA/access-om2/issues/210

[ofa001]

Yes @AndyHoggANU I have identified locations in Gibraltar, Red Sea, Persian Gulf, as well that will need reverting back to at leas the older topography that was used before by the ACCESS-OM2-025 set up, stilll wont resolve the Med getting too salty i long runs that is down to freshwater fluxes incoming into the basin from CABLE, but will be the best we can do.

Still looking at other shallow coasts that aren't confined basins. Will send on plots later this afternoon.

[aidanheerdegen]

I don't know if the model is capable of capturing the natural flows in and our of the Baltic, but here is a description:

Large fresh water surplus and a very limited water exchange with the North Sea maintain the brackish water conditions in the Baltic Sea (HELCOM, 1986). High river discharges in the eastern and northern part and the inflow of high saline water through the Danish Straits establish an estuarine like overturn circulation (Meier et al., 2006). The inflowing dense saline water spreads into the deep layers and causes a strong vertical salinity gradient in the central Baltic with a pronounced halocline at about 60–80 m depth. The inflow of saline water is balanced by upwelling, diapycnal mixing and the outflow of brackish water in the surface layer. Strong stratification at the halocline limits vertical ventilation from the surface to the top mixed layer. Thus, the solely source of deep water renewal and ventilation of the Baltic Sea is the lateral advection of oxygen rich, high saline water.

The water exchange with the North Sea is hampered by the shallow and narrow connections, the Belt Sea and the Sound. The limiting cross sections are the Darss Sill in the Belt Sea and the Drogden Sill in the Sound (Figure 1), with sill depth of 19 m and 8 m, and cross section areas of 612 500 m2 and 52 750 m2, respectively. The flow through these channels is driven by barotropic and baroclinic pressure gradients in the transition area of the Baltic and the North Sea (Lass et al., 1987; Lass, 1988; Feistel et al., 2006). Accordingly, inflow events are divided in barotropic and baroclinic inflows. The latter are driven by the salinity gradient between the Baltic and the North Sea, and occur mainly with calm summer conditions. During the rest of the year the barotropic forcing exceeds the baroclinic forcing considerably. Wind forcing and air pressure differences establish sea level differences between the Kattegat and the western Baltic, which drive barotropic inflow events. To have a significant impact on the deep water conditions in the central Baltic an inflow event must transport large amounts of saline and well oxygenated water into the western Baltic. These events are usually considered as Major Baltic Inflow.

FIGURE 1. Topography of the Baltic Sea and the location of the time series stations used in this study. Gedser Rev (GR), Darss Sill (DS), Drogden Sill (DR), Oskarsgrundet (OG), Kalksgrundet (KG), Bornholm Deep (BD), Gotland Deep (GD), Landsort (LO).

[MartinDix]

More on the Mediterranean.

In ACCESS-CM2-0025, Med + Black Sea total river flow is 1.7e7 kg/s, P-E is -7.1e7 kg/s

From Ludwig et al, https://dx.doi.org/10.1016/j.pocean.2009.02.001, Med river inflow is 1.1 e7, Black Sea is 1.3, so total is 2.4e7 kg/s

From Skliris, https://dx.doi.org/10.1007/s00382-017-4053-7, E-P is 5 - 5.7e7 kg/s

Therefore the model total is -5.4e7 c.f. observed -3.3 to -2.6.

It's bit harder to blame the ocean model inflow past Gibraltar in this case.

[ofa001]

@MartinDix so its the P-E thats the main drive of the salinification, which is also the main driver of the drying trend across Southern Europe so we cant really shift it. But the St of Gibralter is much shallower in the latest 025 topography than in the earlier one COSIMA used than in GEBCO and in what we used in 1deg set up so anything will help. Also I guess the ACCESS- CM2 spin was longer than the ACCESS1-0/1-3 ones thought the ACCESS-ESM1.5 was way longer so that will I expect be much more saline in the Mediterranean.

[ofa001]

@MartinDix The drift in the ACCESS-CM2 piControl and pdcontrol salinities are the outliers, in the ACCESS1-0, the Mediterranean surface salinity was stable at 40psu in year 250, 39 at start of last segment year 745, of 500 year run, also 10psu and 9psu at the start and end in the Baltic. ACCESS1-3 was even more stable 40 psu in the Mediterranean for both segments and 8 for the Baltic for both. The ACCESS-ESM1.5 which had the 2000-2500 plus spin up for the carbon pools has drifted but in 500 year published control is stable 45psu in the Mediterranean though Fresher near the Nile river and Gibraltar , and 5 in the Baltic, so spin up length is also going to be an issue if we have big drifts going into the future as well with long controls.

[aekiss]

I've made some plots of sfc_salt_flux_restore to get a global picture of where the restoring is working hard to keep SSS realistic in ACCESS-OM2 at each resolution. Plot script: https://github.com/aekiss/notebooks/blob/7eff375/salt_restoring.ipynb

This is the time-mean sfc_salt_flux_restore over the six 1958-2018 cycles of the 1 deg run 1deg_jra55_iaf_omip2_cycle[1-6], in units of kg/(m^2*sec). This uses the new GEBCO bathymetry, which has shallower sills in the Baltic, Gibraltar, etc. than in the old OCCAM bathymetry used in ACCESS-CM2.

• Model SSS is too fresh at the mouths of some major rivers (Amazon, Ob, etc). This is a much bigger bias than is seen in the Persian Gulf and Baltic, Mediterranean and Red Seas. If we zoom in the colour scale by 20x we see (click image for a bigger view)
• the Red Sea and eastern Baltic are OK (NB: this was run with redsea_gulfbay_sfix = .false.)
• the Persian Gulf and southern Med are too salty
• northern Med, White Sea and far western Baltic too fresh, with a patch of the southern Baltic that is too salty
• But these are small compared to the river bias and comparable to many of the large-scale open-ocean biases.

On the same colour scale this is the time-mean sfc_salt_flux_restore over the six 1958-2018 cycles of the 0.25 deg run 025deg_jra55_iaf_omip2_cycle[1-6], which uses the same new GEBCO bathymetry as ACCESS-CM2-025.

• Biases are generally smaller than at 1 deg, and we don't have the gross biases at river mouths.
• The Red Sea is now too salty.
• The Persian Gulf and southern Med are too salty, but better than at 1 deg.
• The Adriatic and Aegean are too fresh, as is nearly the entire Baltic and the White Sea.
• There are comparably large biases in many open-ocean and coastal areas.

On the same colour scale this is the time-mean sfc_salt_flux_restore over the three 1958-2018 cycles of the 0.1 deg run 01deg_jra55v140_iaf*.

• Again the Red Sea is too salty.
• The Persian Gulf and southern Med are less salty than before.
• Fresh biases are still present in the Adriatic, Aegean, Baltic and White Seas but most are somewhat smaller than at 0.25 deg.

There have already been edits made to match the old topography in the new 1deg topo and 0.25deg topo files. Comparisons at key straits are shown here.

In summary it looks like these sills could benefit from deepening:

• at 1 deg: Baltic and Persian Gulf - these weren't edited before. The Persian Gulf is already deeper at the mouth than the old topography so it's unclear what to do there. The Med salt bias is unlikely to be improved by editing Gibraltar, as this already matches the old topography, and the restoring indicates that the salty bias doesn't extend into the strait.
• at 0.25 deg: Baltic, Persian Gulf, Red Sea and Gibraltar - these weren't edited before. Gibraltar is shallower than the old topography so copying the old values could help. The Persian Gulf is already very similar to the old topography so it's unclear what to do there.
• at 0.1 deg: Baltic and Red Sea

There are also biases in the White Sea at all resolutions, so maybe something should be done about those too.

We may need to look at the automatic runoff routing to see if rivers are flowing into the wrong embayments (e.g. this issue with the Ob at 0.1 deg: https://github.com/COSIMA/access-om2/issues/130).

[aekiss]

For comparison, here are sfc_salt_flux_restore plots from the runs with old 1deg and 0.25deg topography used in the GMD paper.

This is the time-mean sfc_salt_flux_restore over the five 1958-2017 cycles of the 1 deg run 1deg_jra55v13_iaf_spinup1_B1, in units of kg/(m^2.sec). This uses the old OCCAM bathymetry, the same as in ACCESS-CM2. ...and this is the corresponding plot for 1deg_jra55_iaf_omip2_cycle[1-6] (copied from the previous post), which uses the new bathymetry.

• Overall they are remarkably similar. We are restoring the WOA13 but forcing with JRA55-do, so the patterns may be largely due to a mismatch between WOA13 and E-P in JRA55-do.
• The Baltic, Persian Gulf, Mediterranean and White Sea look very similar. This undermines the argument for editing the Baltic and Persian Gulf sills in my previous post.
• The new topography seems to significantly improve the Red Sea.
• The Amazon outflow looks worse in the new configuration - did the runoff change between JRA55-do v1.3 and v1.4 (used in the old and new configs, respectively)?

This is the time-mean sfc_salt_flux_restore over the five 1958-2017 cycles of the 0.25 deg run 025deg_jra55v13_iaf_gmredi6. This uses the old OCCAM bathymetry. ...and this is the corresponding plot for 025deg_jra55_iaf_omip2_cycle[1-6] (copied from the previous post), which uses the new bathymetry (same as ACCESS-CM2-025). Now we see some significant differences.

• the bias in the Baltic, White and Red Seas changes sign! This suggests there is a sweet spot partway between the old and new topography.
• the magitude of the Baltic bias seems smaller with the new topography. This undermines the argument (in my previous post) for editing the Baltic sills to match the old topography, unless we prefer excessive salinity to excessive freshening. But if we do something in between the old and new depth we may be closer to a sweet spot.
• the magitude of the Red Sea bias is worse with the new topography, supporting the argument for deepening its sill.
• the salty bias in the Mediterranean and Persian Gulf is worse with the new topography, supporting the argument for deepening their sills.
• the Black Sea is much better with the new topography (or JRA55-do version?)
• as for the 1deg case, the Amazon outflow looks worse in the new configuration (which again switched from JRA5-do v1.3 to v1.4)

[ofa001]

Thanks @aekiss for these plots and apologies for the delay in replying, The Black Se a was definitely better in the new topography, the Baltic, Mediterranean sill, Red Sea were closer in the older one, the Gulf is fairly flat so not much of a sill, but the old was a little deeper at the sill. Interesting if it will be a compromise between the depths in the two topography cases.

Have also looked at Indonesian region and Torres St which have a shallowed a bit not sure they are as critical, though it may impact sea level rise calculations there if we use the 12m minimum depth there, but it is very shallow in reality.

[aekiss]

More detail on the White Sea at 0.25 deg:

new: 20m sill depth

old: 103m sill depth

difference (blue=shallower in new).

SSS restoring changes sign, suggesting there is a sweet spot partway between the two.

gebco: sill is about 40m deep, so that would be good to try.

[aekiss]

More detail on the Persian Gulf at 0.25deg:

new: 51m sill depth

old: 71m sill depth

difference (blue=shallower in new)

gebco sill depth looks closer to old topo:

[aekiss]

@ofa001 mentioned concerns about the eastern Bering Sea and Gulf of Alaska (emails 20, 21 Oct) so here are some views of that region.

new 0.25deg topography:

old 0.25deg topography:

difference (blue=shallower in new):

gebco shows that the pits on the shelves in the old bathymetry are not realistic (see https://github.com/mom-ocean/MOM5/issues/172) and the new topography is closer to reality (not surprising, since it's based on gebco)

[ofa001]

Thanks @aekiss for those new plots @MartinDix pointed out that the high salinity values in the shallow part of the Gulf of Alaska reached a maximum and didn't keep increasing, so hopefully wont keep increasing, like they do in the high evaporation regions of the Persian Gulf and Red Sea, but you are right the deep pits in the old topography there were unrealistic just the high salinity is now an obvious strange point, but we can live with if it doesn't cause model problems.

The White Sea and Perssain Gulf sills it may be worth running with the original sill depths based on your salinity relaxation data.

[aekiss]

@ofa001 also pointed out (email 29 Oct) areas against the Antarctic coast in the 0.25deg topography with depth 11.8m (ie the min depth), which coincided with velocity blowups in @MartinDix's run.

new 0.25 topo, showing 11.8m-deep regions

old 0.25deg topo, showing similar regions, but at the old min depth of 40m:

difference (blue=shallower in new):

also for the east Antarctic

new, showing shallow patches in eastern Ross and a few other places:

old, with more extensive shallow regions

difference, showing new topog is deeper at coast but shallower offshore in Amundsen-Bellingshausen Sea This was noted above

gebco looks like this

Perhaps the shallow parts are due to a misalignment of the gebco and model land masks, i.e. land in gebco, ocean in model, e.g. the peninsula on the ~western~ eastern Weddell Sea.

The SSS restoring doesn't look very different between old and new topography, so if this were to be fixed it would be for velocity stability reasons, not SSS.

[aekiss]

OK @MartinDix, @ofa001, @AndyHoggANU, @aidanheerdegen here's my attempt to summarise the discussion starting here and via email regarding improvements to the 0.25deg topography:

1. Baltic Sea - SSS restoring changes sign with the new topog, suggesting there is a sweet spot partway between the old and new topography. The magnitude of the Baltic bias seems smaller with the new topography - so set to partway between 11.8 and 40m - e.g. 19m to match Darss Sill.
2. Gibraltar - Siobhan (email 22 Oct) suggests at least 250m, based on gebco. Sill depth in new topog is 70m; in old topog it's 288m, which matches the Camarinal Sill depth (284m). So set to match old topog.
3. Red Sea - SSS restoring changes sign, suggesting there is a sweet spot partway between the old (138m) and new (15m sill) topography. The magnitude of the Red Sea bias is worse with the new topography, suggesting its sill should be deepened to be close to the old topography. The Bab-el-Mandeb sill is 137m deep so we should match the old topography. (Appropriately enough, Bab-el-Mandeb means "Gate of Lamentation", so I should have taken heed and paid it more attention...)
4. Persian Gulf - The Persian Gulf is very similar to the old topography but it's a little shallower near the mouth (the Strait of Hormuz) in the new topog (old depth is 71m, new is 51m) so we could at least revert to the old topog there as it seems more consistent with GEBCO and a few other observational depth values I could find.
5. White Sea - SSS restoring changes sign, but has roughly the same magnitude, suggesting there is a sweet spot partway between the old (103m sill) and new (20m sill) topography. The sill at Gorlo Strait is about 40m deep so we could try that.
6. Antarctic coast - might be a land mask mismatch. Velocity blowups in shallow regions on eastern Weddell Sea. This could take some time and discussion to work out what to do, so don't modify topog for the first iteration.
7. eastern Bering Sea - salinity high but steady, and new topography is more realistic so don't modify topog - see emails 20, 21 Oct
8. Gulf of Alaska - Siobhan emails 20 Oct - don't modify topog. See here.

I'd like to make a start on this soon (maybe tomorrow), so please let me know if I've missed anything or you disagree. This is only a first iteration for testing with ACCESS-OM2-025 prior to trying it with ACCESS-CM2-025.

[AndyHoggANU]

Aah, the good old gate of lamentation.

Yes, this all looks good to me. I'm happy to run the OM2 test case once we have a sample topo -- @MartinDix, once I can get far enough into that to know that areas 1-5 have improved then it is probably worth deciding either (a) further optimisation is needed; or (b) we should test in CM2.

[ofa001]

Hi @aekiss, I think I would go with your list of suggestions, the only one of concern still is the Antarctic coastline one where you think it could be a land mask match do you think that might be a land mask mismatch in the coupled model, or with the GEBCO data, the UM will have those cells as partial land/ocean cells so will treat the wind stress as a mixture of land and ocean/ice. We can discuss more. The shelf is quite wide though its not just 1 or two points.

I also noted the "gate of lamentation' name, I guess that must have come from some Arab sailors.

[aekiss]

@AndyHoggANU I've made an updated 0.25deg topography addressing items 1-5 above: /g/data/ik11/inputs/access-om2/input_20211111_025deg_topog/mom_025deg/topog.nc

We decided you could do a test as a cold-start from the start of omip cycle 1. This is commit cc03806 on branch omip_amoctopo_cycle1 https://github.com/COSIMA/025deg_jra55_iaf/tree/omip_amoctopo_cycle1 The run history is here https://github.com/COSIMA/025deg_jra55_iaf/blob/omip_amoctopo_cycle1/run_summary_home_561_rmh561_access-om2_025deg_jra55_iaf_amoctopo_cycle1.csv

[aekiss]

The updated topography /g/data/ik11/inputs/access-om2/input_20211111_025deg_topog/mom_025deg/topog.nc was generated by https://github.com/COSIMA/make_025deg_topo/tree/9541dc1, which used these edits.

For the sake of documentation, here's a whole lot of plots showing the changes. The locations of the edited points are marked with red circles.

Baltic Sea - set to 19m depth (item 1 above): before edit: after edit: change (red is deeper in new version): old OCCAM topography: difference from OCCAM topography (red is deeper in new version):

Gibraltar - match to old OCCAM topography (item 2 above): before edit: after edit: old OCCAM topog: difference from old OCCAM topog before editing: difference from old OCCAM topog after editing:

Red Sea - match to old OCCAM topography (item 3 above): before edit: after edit: change (red is deeper in new version): old OCCAM topog: difference from old OCCAM topog before editing: difference from old OCCAM topog after editing:

Persian Gulf - match to old OCCAM topography (item 4 above): before edit: after edit: change (red is deeper in new version): old OCCAM topog: difference from old OCCAM topog before editing: difference from old OCCAM topog after editing:

White Sea - set to 40m depth (item 5 above): before edit: after edit: change (red is deeper in new version):

old OCCAM topog:

difference from old OCCAM topog after editing:

[aekiss]

re. the Antarctic coast (item 6 above), if the model has ocean where gebco has land (actually ice shelves), the automatic topog generator will set it to the minimum depth, which is 11.8m in the 0.25deg case. This needs further investigation. We may need to provide more realistic values in these regions, e.g. from MEaSUREs BedMachine Antarctica, Version 2 - see https://coecms.github.io/2021/11/02/pyproj-regrid.html

[ofa001]

Yes @aekiss the values closest to the coast could well be ice shelves, but they often sit in quite deep water, so its GEBCO that put it in shallow ones I guess there is things like BEDMAP which is probably the basis for the dataset you are referring to above. that has all the Antarctic topography and ice sheet in it but its not on a nice lat/lon grid. Don't know if you have seen the CM2 0.25 output but the salinities got > 40 psu values in winter I guess from brine rejection, and then getting trapped. So yes we will have to come up with a solution, perhaps we should look at HADGEM and GFDL 0.25 grids which I meant to do for some other locations.

[ofa001]

Thanks @aekiss for documenting those topography changes before the new runs start.

[AndyHoggANU]

@aekiss - a quick update for you here. I have looked at the output for the first 20 years of the test with the new topography. I will keep running it overnight and check this more carefully in the morning, at which stage I will send you some details. But my preliminary findings are:

• Red Sea and Med are markedly improved with the new topo.
• White Sea and Persian Gulf are probably better, certainly no worse.
• Baltic is almost unchanged. Based on this, I think we may need to consider doing another 2 gridpoints -- to 31 m -- in the centre of the strait there. Details to come ...

[ofa001]

Thanks @AndyHoggANU yes it sounds like the Baltic is the crucial one as that where it reached to zero salinity in the coupled model, so increasing the topography a bit more may be needed in the OM tests if the current changes haven't increased the flow.

[aekiss]

Thanks @AndyHoggANU - where's the output for this run? I'd like to take a look.

[AndyHoggANU]

No worries - the run is at /g/data/ik11/outputs/access-om2-025/025deg_jra55_iaf_omip_straits_topo_cycle1. I will put in a few figures here to explain what I talked about last night. The run has gone further than this, but I am struggling with databases right now.

I said that Med is looking good. By this I meant that salinity has dropped and restoring approaches zero. See this figure, which is an average of salinity over a large box in the Med, and a sum of freshwater flux from salinity restoring:

Note that this compares Ryan's original OMIP run (in blue) with the new TOPO run (orange).

[AndyHoggANU]

Red sea is similar to the Med:

[AndyHoggANU]

And here is the Baltic. Hardly changes at all .

[AndyHoggANU]

I'm not convinced the Persian Gulf or White Sea are better, but they are probably not worse ...

[AndyHoggANU]

Here I would generally assume that the initial state is a good salinity to aim for (because it is WOA) and that zero salinity restoring would be ideal...

[aekiss]

Thanks @AndyHoggANU, I agree Med and Red are looking much better now (though Med salinity has dropped relative to the WOA initial condition - is this SSS or a depth-average?)

[russfiedler]

That massive drop of salinity in the Persian gulf in the 70's looks suspicious. Huge increase in rainfall or river output. Is it misplaced?

I think you need some transects of salt transport for the Baltic just to see if you've been able to generate any sort of inflow from tha Atlantic. Maybe Oresunde to the east of Zeeland needs to be opened (or xland mixing in the bottom few cells).

[aekiss]

Thanks @russfiedler, that's a good point re. Oresunde but I'm reluctant to change the land mask to open up that channel, as it usually plays a secondary role in salt inflow due to the shallowness and narrowness of Drogden Sill relative to Darss Sill (see above: The limiting cross sections are the Darss Sill in the Belt Sea and the Drogden Sill in the Sound (Figure 1), with sill depth of 19 m and 8 m, and cross section areas of 612 500 m2 and 52 750 m2, respectively).

But as an engineering fix it is certainly a more direct route, so we should bear than in mind in case deepening/widening the southern channel doesn't fix things.

[russfiedler]

Yeah, I saw that in Aidan's post above which is why I mentioned xland transport as it can somewhat mimic estuarine flow if needed. A quick looks indicates there is a little bit of estuarine style flow to the west of Zeeland but virtually nothing makes it into the central Baltic.

[ofa001]

Just been looking at the GFDL-CM4 output from nci's oi10 space, they have both channels open in the Baltic entrance but are on a C grid so can manage the velocity points better However in the vertical they only have 2 points 2.5 and 10m. Also have only 1 psu salinity by then end of 500yr control. In the Gulf and Northern Red sea they get very high salinities will email some plots shortly.

[AndyHoggANU]

@aekiss - my salinity metric is a pseudo depth average (it's just a xarray mean, so not properly weighted, I am afraid). I didn't want to use surface as that is more infected by the restoring. I can share my notebook if you want to look?

Is there anything else you want to look at with the next version of the topo? I am happy to put on a case with the new Baltic when it is ready -- I think we will only need 10 years to know it has worked...

[aekiss]

I've updated the topography /g/data/ik11/inputs/access-om2/input_20211116_025deg_topog/mom_025deg/topog.nc with https://github.com/COSIMA/make_025deg_topo/tree/67a76dd, which used these edits.

The outlet to the Baltic Sea has been deepened from 19 to to 31m and widened to 3 cells (ie 2 velocity points). before: after: difference from OCCAM (blue is shallower than OCCAM):

More cells in the Persian Gulf outlet have been deepened to match OCCAM (might not make much difference). before: after: difference from OCCAM (blue is shallower than OCCAM):

The White Sea outlet has been deepened from 40m to 70m (possibly overdid it, but we'll see). before: after: difference from OCCAM (blue is shallower than OCCAM):

[aekiss]

Here's the salt flux restoring averaged over 1958-1985 for Andy's first test /g/data/ik11/outputs/access-om2-025/025deg_jra55_iaf_omip_straits_topo_cycle1 which used /g/data/ik11/inputs/access-om2/input_20211111_025deg_topog/mom_025deg/topog.nc:

For comparison, here's the OMIP run /g/data/ik11/outputs/access-om2-025/025deg_jra55_iaf_omip2_cycle1 which used /g/data/ik11/inputs/access-om2/input_20201102/mom_025deg/topog.nc (as in CM2-025), averaged over the same time 1958-1985 period:

It looks like the Med and Red seas are about as good as we can expect (unless we want to improve the Adriatic, though this might be related to runoff), but more work is needed on the Persian Gulf, Baltic and White seas, as Andy concluded above. This informed the choices I made in the 2nd iteration to the topography.

---

EDIT 22 Nov Here's the salt flux restoring averaged over 1958-1985 for Andy's 2nd test /g/data/ik11/outputs/access-om2-025/025deg_jra55_iaf_omip_straits_topo2_cycle1 which used the 2nd iteration topography /g/data/ik11/inputs/access-om2/input_20211116_025deg_topog/mom_025deg/topog.nc:

Now the Persian Gulf and White Sea look pretty good, but the Baltic is almost unchanged relative to the first adjustment (top figure), with the only noticeable change being a small increase in the blue (ie excessive SSS) region on the eastern side of the straits.

For reference here's the salt flux restoring averaged over 1958-1985 (starting from WOA) for 025deg_jra55v13_iaf_gmredi6 (used in the GMD paper), which used the OCCAM-based topography and JRA55-do v1.3 instead of 1.4: The new topography improves the Med, Red, Black and White seas and is not much worse in the Persian Gulf. The Baltic is better in magnitude but the opposite sign. It is odd that this is so different. Perhaps the JRA55 runoff is different in the two versions.

Also here's the salt flux restoring averaged over 1958-1985 (starting from WOA) for 1deg_jra55v13_iaf_spinup1_B1 (used in the GMD paper), which used the OCCAM-based topography and JRA55-do v1.3 instead of 1.4. This is the same topography as in ACCESS-CM2. This has somewhat less SSS bias in the Baltic than our latest topography, but is much worse in the Baltic straits and all the other seas.

[aekiss]

The Antarctic issue (item 6 above) also needs to be fixed. This is due to the topogtools/apply_mask.py step in the make_topog.sh processing pipeline, which sets ocean cells that are land cells in GEBCO to the minimum depth of 11.8m. This is a reasonable thing to do on continental coasts where the ocean can be expected to be shallow, but is a terrible idea off the Antarctic ice shelves where the ocean is hundreds of metres deep.

The top plot (click to zoom in) shows the topography immediately after apply_mask.py, and the middle plot shows the topography immediately before this step, when the land mask is still that from GEBCO. apply_mask.py produces the very shallow (brown) regions in the top plot in the eastern Ross, eastern Antarctic Peninsula, southwestern Weddell, east of Weddell, Prydz Bay and elsewhere.

So the question is what to do about it. Options include:

1. replacing the filled values with something more reasonable than the minimum depth, e.g. 1a. using the old OCCAM-based values from /g/data/ik11/inputs/access-om2/input_20200530/mom_025deg/topog.nc (the bottom plot); or 1b extrapolating from GEBCO; or 1c using MEaSUREs BedMachine Antarctica
2. changing the model land mask to match GEBCO

Option 1a seems a reasonable choice to me, and is also the easiest to do.

[PaulSpence]

Hi Folks,

FYI: the bedmachine v2 topography has been regridded (from x/y in metres to lat/lon) onto the ACCESS-om2-01 grid here: /g/data/v45/pas561/bedmachineant/bathy_om2.nc (supreme thanks to Aidan).

Thank you, Paul

On Wed, Nov 17, 2021 at 12:45 PM Andrew Kiss @.***> wrote:

The Antarctic issue (item 6 above https://github.com/COSIMA/access-om2/issues/158#issuecomment-963889409) also needs to be fixed. This is due to the topogtools/apply_mask.py https://github.com/COSIMA/topogtools/blob/cca23d9/apply_mask.py step in the make_topog.sh https://github.com/COSIMA/make_025deg_topo/blob/67a76ddc/make_topog.sh#L37 processing pipeline, which sets ocean cells that are land cells in GEBCO to the minimum depth of 11.8m. This is a reasonable thing to do on continental coasts where the ocean can be expected to be shallow, but is a terrible idea off the Antarctic ice shelves where the ocean is hundreds of metres deep.

The top plot (click to zoom in) shows the topography immediately after apply_mask.py, and the middle plot shows the topography immediately before this step, when the land mask is still that from GEBCO. apply_mask.py produces the very shallow (brown) regions in the top plot in the eastern Ross, eastern Antarctic Peninsula, southwestern Weddell, east of Weddell, Prydz Bay and elsewhere. [image: Screen Shot 2021-11-17 at Wed 17-11 12 20pm] https://user-images.githubusercontent.com/31054815/142092263-02cff740-5179-4fc9-ab00-293d842d8afa.png

So the question is what to do about it. Options include:

1. replacing the filled values with something more reasonable than the minimum depth, e.g. 1a. using the old OCCAM-based values from /g/data/ik11/inputs/access-om2/input_20200530/mom_025deg/topog.nc (the bottom plot); or 1b extrapolating from GEBCO; or 1c using MEaSUREs BedMachine Antarctica https://github.com/COSIMA/access-om2/issues/158#issuecomment-966666298
2. changing the model land mask to match GEBCO

Option 1a seems a reasonable choice to me, and is also the easiest to do.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/COSIMA/access-om2/issues/158#issuecomment-971064268, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABSWJXEAC3TFTYXX4G7QOL3UMMCMFANCNFSM4IL237RA . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

-- I’m sending this message at a time that suits me. I don’t expect you to reply outside of your own work hours.

Paul Spence, Assoc. Prof. ARC Future Fellow Institute for Marine and Antarctic Studies https://www.imas.utas.edu.au/ University of Tasmania, Hobart, Australia https://paulspence.github.io/

[adele-morrison]

My intuition is that

2. changing the model land mask to match GEBCO would be the most sensible choice. These locations that are land in GEBCO all seem to be ice shelves, and elsewhere around Antarctica we set those to be land in the model.

Is this also a problem for the 1/10th degree bathymetry?

It looks like the ocean is being cropped at the most southerly latitudes also. In particular, the eastern Ross Sea shelf has about 1/2 cut off. In the 1/10th degree we extended the southern most limit of the model further south to fix this.

[aekiss]

This issue doesn't affect the 0.1deg topography. @russfiedler created the 0.1deg topography via a different workflow (not using apply_mask.py), and was starting from scratch with GEBCO and appears to have used the GEBCO land mask: so perhaps that would be a good way to go, so the land masks at 0.25 and 0.1deg would match, also fixing the cutoff of the Ross and Weddell seas. There are also significant land mask differences in the Arctic but I'm reluctant to expand the scope any further than we need to.

[ofa001]

It might be worth talking to @MartinDix on how easy it would be alter the land/ice sheet mask and its orography though at the coast these are ice shelve so are not much above sea level.

@aekiss is right option 1a is easiest but consistency with 1/10 degree model might be more sensible.

[MartinDix]

Updating the atmosphere mask is pretty simple so shouldn't influence the choice. This was the region of a few of the model crashes we had so might improve the stability too.

[aekiss]

The GEBCO land mask will need some tweaking if there are non-advective points.

[ofa001]

@aekiss There looked like there was an isolated point over in the corner of the Ross Sea even in this mask, but it might have been 2 tracer points I will need to check it went to a very high salinity in the coupled run.

[aekiss]

@ofa001 non-advective points have already been detected and removed from the current topography (top panel above) - see https://github.com/COSIMA/access-om2/issues/210

[ofa001]

Thanks @aekiss

[aekiss]

If we extend the ocean too far southward in the Ross and Weddell we'll need to activate some of the inactive tiles and increase the core count

[aidanheerdegen]

Is that a problem?

[aekiss]

no, just more to do