Options for allocating roots within active soil layers or using dynamic roots in FATES?

[huitang-earth]

Hi, CLM has the options to allocate roots within unfrozen or active soil layers in the permafrost region ("perchroot", "perchroot_alt"). It also has an option for using dynamic roots ("use_dynroot"). These features seem to have large impacts on vegetation dynamics in the boreal and Arctic region. I am wondering if these features could or will be included in FATES in the future? Hui

[ckoven]

@huitang-earth good question. I don't think this is on anyone's to-do list yet. Agreed that, in permafrost regions, the root profile used for btran calculations should probably work like it does in CLM, where it truncates and renormalizes the root profile used in computing the btran such that it is zero below the active layer but still has the same vertical integral as the full profile. And not sure how this would work in hydro but should probably use similar logic.

Given the structural differences between FATES and CLM, a way to accomplish this could be to restrict the rooting profile used in the btran calculation to only go from depth index 1 down to bc_in(s)%max_rooting_depth_index_col. Because the max_rooting_depth_index_col variable keeps track of the active layer depth, doing so should effectively prevent roots from growing into the permafrost (and getting zero moisture once there).

[ckoven]

Its also possible that this is already happening and I'm just not perceiving it in the code. @rgknox just rewrote much of the root profile code to handle CLM5's variable soil depth logic so I am possibly out of date in my reading of the code.

[rgknox]

Thanks for bringing this up @huitang-earth . FATES will specify that its roots will go as far down as the bedrock (depth of which is dictated by CLM/ELM). The rooting depth parameter will control how much of the root's normalized profile will approach that depth, but it will be cutoff at bedrock, perhaps abruptly. As Charlie mentions, we will also ask CLM/ELM what the active layer depth is, "bc_in(s)%max_rooting_depth_index". That layer will only set the depth, over which, litter fluxes are passed from live roots to the soil decomposition pools, and this depth will be either as deep as or shallower than the bedrock depth. So if this is shallower than the bedrock depth, we will have a fraction of the root tissue that is effectively frozen in the soil, somewhat intert... At least that is how I am interpreting what is going on.

The relevant parts of the code for this question seem to start here:

https://github.com/NGEET/fates/blob/master/biogeochem/EDPhysiologyMod.F90#L2137

[ekluzek]

All, just a note, that use_dynroots isn't working well scientifically and not recommended for use. It's going to require more science development before people should use it. It was intended to be a core part of CLM5.0, but there were issues with it and it can only be turned on as an experimental option.

[rosiealice]

I'd guess that HYDRO would actually treat this completely differently since it would calculate whether it can actually extract water from a layer, .given the potentials and resistances in that layer, and so where the roots actually are in the soil will have less of an impact on the overall water stress...

Also, yes, as @ekluzek mentioned, dynamic roots proved difficult to implement in CLM5 (mostly as far as I understand in the sense that it was very very hard to equilibrate).

[ckoven]

I looked at this a bit more.

For the non-hydro code, the question is whether the btran is calculated in either seasonally frozen or permanently frozen soils. There is logic in the btran code that ensures that btran is only calculated over currently-unfrozen soil layers, here: https://github.com/NGEET/fates/blob/master/biogeophys/EDBtranMod.F90#L142 So that is equivalent to the CLM logic.

For the hydro code, I don't see any equivalent logic. So I'm a little concerned about how this would behave in terms of cryosuction. In the soils, when the temperature of a given layer is below freezing, the liquid water potential there becomes strongly negative, so the moisture wants to flow along a temperature gradient from warm to cold (which is a real thing). But the hydraulic conductivity in those frozen layers is also very low, so that in practice you don't get all that much transport of water. But in the hydro code, the root hydraulic conductivity doesn't appear to have any freeze inhibition, or even include the effect of freezing on plant tissue psi. So I'm sort of afraid that where the roots touch a frozen soil layer, they will have the water sucked out of them, and then also possibly serve as a conduit for sucking the water from unfrozen soil layers and plant tissues. Am I reading the hydro code right, or is there a safeguard here that I'm not catching? cc @bchristo @xuchongang

[bchristo]

@huitang-earth @ckoven @xuchongang

Indeed, Charlie is correct -- the current FATES-Hydro code has no safeguard against this, and the code needs to be adapted to account for freezing. Quite frankly I am not too familiar with water flow under freezing conditions, but it seems to me the simplest way to address this is to apply the same HLM approach to freezing (via modification of hksat or k) to the hydro conductivities as well.

[ckoven]

ok another addition to FATES's to-do list!