Add new feature: Lateral flow of Dissolved Organic Carbon (DOC)

[devarajun]

To begin with, solving the mass balance equation for DOC transport is planned. This will imply changes in MOSART (https://github.com/ESCOMP/MOSART/blob/master/src/riverroute/MOSART_physics_mod.F90)
and in CTSM Soil Carbon Module (https://github.com/ESCOMP/CTSM/blob/master/src/biogeochem/CNVegCarbonStateType.F90).

Brief science information:

Assume the lateral transport of DOC to be proportional to the water fluxes, i.e. the export from one reservoir to the next have the same concentration of DOC. DOC is drawn up into runoff from the top 5cm of soil layer.

At each spatial unit (lat/lon grid), DOC goes into water fluxes generated from hillslopes, then tributaries and to main channel. DOC implementation code mainly builds upon on MOSART code for taking up DOC into hillslope, tributaries and main channel.

Plan for the implementation of the DOC mass balance code:

We plan to implement the changes for Soil Carbon module and MOSART separately in their respective repositories.

Pseudocode for MOSART

DOC mass balance equation mainly requires the total water storage for each category of hydro-logical unit and the channel flow rate. This information used in the mass balance equation of DOC transport.

Pseudocode for CTSM Soil carbon module

DOC production in soils derived from the first-order kinetics formulation for different carbon pools (see the article for the DOC production formula https://gmd.copernicus.org/articles/11/593/2018/).

Long term goal

• Long term plan is to couple the microbial soil decomposition model (developed by Elin Ristorp Aas, PhD Student at University of Oslo) with MOSART-DOC.

• If possible to extend simple DOC mass balance to follow Liao et al (2019) https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2019MS001792

People involved: @devarajun , @sunnivin , @ecaas (Elin Ristorp Aas) @kjetilaas (Kjetil Schanke Aas)

[wwieder]

Thanks for filing this issue, @devarajun. Looking forward to seeing your contributions down the road.
It may be worth coordinating the efforts to integrate the land changes to soil BGC with mizuRoute #920 instead (or in addition) to MOSART? Also, I'm assuming you'll produce DON fluxes with the DOC fluxes? We may as pass these to the river model too?

[devarajun]

Thank you @wwieder for commenting. Regarding DON, yes we have plans for DON fluxes. We were not aware of mizuRoute and thanks for pointing this out. How different mizuRoute from MOSART in terms of biases in high latitudes? I will try to explore mizuRoute documentation. Is it already a optional component of CTSM/CLM and works in regional CLM simulation setup at any resolution?

[wwieder]

I think don't think mizuRoute is ready to go yet, but an upcoming feature that's being actively developed now. @nmizukami and @ekluzek can you confirm? @devarajun I mainly wanted to give you a heads up that mizuRoute is coming. I trust adding tracer fluxes aren't an overwhelming burden in either river model.

[dlawrenncar]

Minor point, but should stress that we don't really expect streamflow biases to be strongly affected by using MOSART or mizuRoute, especially for big rivers. The primary source of the biases in most cases, I think, is the amount of runoff that is generated by CTSM. There are other benefits to mizuRoute beyond reducing biases though, including better integration between rivers and lakes and a more accurate river network, especially for smaller rivers.

…

On Thu, Nov 19, 2020 at 5:29 AM will wieder ***@***.***> wrote: I think don't think mizuRoute is ready to go yet, but an upcoming feature that's being actively developed now. @nmizukami <https://github.com/nmizukami> and @ekluzek <https://github.com/ekluzek> can you confirm? @devarajun <https://github.com/devarajun> I mainly wanted to give you a heads up that mizuRoute is coming. I trust adding tracer fluxes aren't an overwhelming burden in either river model. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#1216 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AFABYVG4BAGDNWSSVVBLJRLSQUFUHANCNFSM4TYZ2Z3A> .

[ekluzek]

@wwieder we do have a preliminary version of mizuRoute that comes with CTSM as of ctsm5.1.dev016. It doesn't really function there though, but you can at least look at a version of the code. I have a branch on my fork add_mizuRoute that we have working for a few configurations. We'll likely only have mizuRoute working with CTSM for a limited set of resolutions.

Also mizuRoute isn't currently handling tracers. @nmizukami, and @martynpclark will have to let us know about how hard that will be to add. As @dlawrenncar points out it might make the most sense to add certain features to only one river model (especially at first) for the configuration that it makes the most sense in. Right now it seems like the roles are: RTM will be the river model for Paleo work, MOSART for climate, and mizuRoute for NWP. Dynamic lakes is a feature that's just going into mizuRoute, so we already have precedence for features going into one and not all river models. And practically you can't always put all needed features into each of the different river models. I imagine in the long term it's possible mizuRoute will replace both RTM and MOSART, but right now we need all three.

[nmizukami]

Hi @devarajun, adding to comments from @ekluzek, the main difference between mizuroute and mosart/rtm is river network. mosart/rtm use gridded network while mizuRoute uses catchments (you can think this as unstructured mesh, but complex shape). The difference in river network can change river discharge if drainage area is different.
For tracer capability, I think we (including @martynpclark) will need to discuss the best path forward.

[ekluzek]

Another aspect of this is that DOC will need to be passed from CTSM through the coupler to MOSART. We'll likely want to implement this in the NUOPC coupler CMEPS rather than the current MCT coupler. There might also need to be an XML variable to trigger that this should be turned on in both CTSM and MOSART. This would likely be similar to this issue for the support of mizuRoute...

ESCOMP/CMEPS#132

[mvertens]

@ekluzek - thanks for pointing out the implementation in CMEPS. I am happy to help with this and review the implementation.

[devarajun]

Thanks to all of you for your useful inputs. @ekluzek great that you have pointed out on passing DOC through the coupler and I now know this at the beginning of coding thanks! @nmizukami @mvertens will keep in touch with you.

[devarajun]

Hi, @ekluzek @nmizukami @mvertens @wwieder @dlawrenncar
As I explained earlier my intention was to implement a subroutine directly into MOSART. However, considering mizuRoute (and RTM) river routing options you have mentioned, if its feasible to make DOC/DON work with any current and future river routing. What do you think is the best way of implementing DOC/DON code? It would be straight forward to include a DOC/DON subroutine in either of the river routing components. But with respect to future developments and compatibility, I wonder if creating a module or subroutine for DOC/DON in CLM biogeochem source code that access stream-flow in & out and runoff through the coupler from either MOSART or mizuRoute, could be a better choice? This would require or at least benefit from a well defined interface to river routing sub-components. By the way, the mass-balance equation of tracers along with the source term given below mainly requires lateral flow in and out for each hydrological unit and runoff generation (for eg. hillslope, sub-network, and the main channel for MOSART). Q→ Streamflow, c→ concentration of DOC/DON, r→ DOC source/sink term.
d( CS)/dt = sum(QC)_inflows - sum (Q*C)_outflows + r

[nmizukami]

It seems to me that it would not be very difficult to implement the tracer based on mass balance equation like that (advection only) into river models (without any details in each river model).

I wonder if creating a module or subroutine for DOC/DON in CLM biogeochem source code that access stream-flow in & out and runoff through the coupler from either MOSART or mizuRoute, could be a better choice?

Your routine needs to know flow direction to move DOC/DON downstream at land model grid? IMO: DOC/DON is passed from land model to river, and the tracer subroutine should reside in the river model component.

Which river models should be used? maybe others can weigh in?

[dlawrenncar]

For now, I think the main river model should be MOSART with the goal of also eventually being able to use mizuRoute. I don't know enough to be able to anticipate whether or not it would be possible to make this general enough to be applicable across multiple river models. That will take some thought and exploration. Perhaps we should meet after the holidays to discuss.

…

On Tue, Dec 22, 2020 at 12:08 PM Naoki Mizukami ***@***.***> wrote: It seems to me that it would not be very difficult to implement the tracer based on mass balance equation like that (advection only) into river models (without any details in each river model). I wonder if creating a module or subroutine for DOC/DON in CLM biogeochem source code that access stream-flow in & out and runoff through the coupler from either MOSART or mizuRoute, could be a better choice? Your routine needs to know flow direction to move DOC/DON downstream at land model grid? IMO: DOC/DON is passed from land model to river, and the tracer subroutine should reside in the river model component. Which river models should be used? maybe others can weigh in? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#1216 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AFABYVBKP6CAK6H7N37C7KTSWDVEPANCNFSM4TYZ2Z3A> .