Components of water balance at the plot level

[Manuela-RT]

Dear all,

I would like to calculate the water balance components at the plot level in terms of the general formula Precipitation = ET + Runoff + Change in soil storage using the monthly outputs from an ED 2.2 simulation.

After reading the reference Longo et al., (2019), I understand that ED 2.2 calculates the water balance for each thermodynamic system, i.e., each soil layer, TSW, cohorts, and CAS. After checking in detail each component in the net water flux equations for each of these systems, I am not sure if I got correctly all the components I need in order to calculate the water balance in the more general form (Precipitation = ET + Runoff + Change in soil storage) at the plot level.

So far, I identified these components in my ED 2.2 outputs: rainfall (from meteorological drivers), ET from each cohort, evaporation from the soil, runoff (surface and subsurface) and change in soil moisture. However, the sum of the components do not correspond exactly to the rainfall inputs. Is there something I am missing? Also, I am not sure if I identified correctly the name of each variable in the ed_state_vars.F90

This is something I would like to work with for the last chapter of my PhD thesis and I'd appreciate any help.

Regards,
Manuela (m.a.rueda.trujillo@cml.leidenuniv.nl)

[mpaiao]

@Manuela-RT From the top of my head, I can think of the following:

1. For leaf ET, which variables are you using? The total vegetation ET should be the sum of three variables:

• mmean_transp (transpiration)
• mmean_vapor_lc (evaporation from the leaf surface)
• mmean_vapor_wc (evaporation from branches). This may or may not be zero depending on the NL%IBRANCH_THERMO you used.

2. The total water storage is technically the combination of multiple pools:

• soils, which you are already considering, and by far the largest pool.
• temporary surface water (mmean_sfcw_mass). In the tropics it is typically short-lived, as there is no snowpack, but it can be non-zero during rainy periods
• canopy air space: this is much smaller than the soil storage, but again, it is non-zero. Unfortunately I don't think we have a variable that has the mean canopy air space storage, but you can get a close estimate in kg H2O / m2 by defining: mmean_can_storage = mmean_can_rhos * mmean_can_shv * can_depth)

If you don't mind, could you please post the calculation you are doing, so we can take a look and see what may be missing?

[Manuela-RT]

@mpaiao These are the variables I am using at the moment:

Precipitation = MMEAN_PCPG
ET = MMEAN_VAPOR_LC + MMEAN_TRANSP + MMEAN_VAPOR_GC (NL%IBRANCH_THERMO set to zero)
Change in soil storage = MMEAN_SOIL_WATER (the difference month to month)
Runoff = MMEAN_RUNOFF + MMEAN_DRAINAGE

Regarding the variable MMEAN_SFCW_MASS you suggest, I thought -without being sure- that in the tropics, where temporary surface water is short-lived, it probably ends either as evaporated water from the soil, in the soil water storage, lost to runoff or utilized by plants at the monthly scale, so I did not want to "double" count it. Do you think I should include it too?

I'll try to calculate the change in canopy air space storage as you suggested.

Thank you!

[mpaiao]

@Manuela-RT I think you must include MMEAN_SFCW_MASS as part of the storage. You are absolutely right that this pool is short lived. However, if MEAN_SFCW_MASS is non-zero, then it means that temporary surface water stored water at some point. If you don't include it, you may underestimate storage, which could cause lack of water conservation.

If you do include change in canopy air space storage, you may want to replace ET with the eddy flux at the top of the canopy air space (−1 × MMEAN_VAPOR_AC). That said, I would expect some small departure from perfect closure at monthly scale because the storage in canopy air space should really be:

MMEAN(CAN_RHOS × CAN_SHV × CAN_DEPTH)

instead of

MMEAN_CAN_RHOS × MMEAN_CAN_SHV × CAN_DEPTH

I think in the future we should have an output variable that reports the enthalpy and water storage in the canopy air space...

[xiangtaoxu]

functions in budget_utils.f90 might be a good reference too. For example, the compute_water_storage function demonstrates all water pools that need to be considered

ED2/ED/src/utils/budget_utils.f90

Line 1373 in bae4504

real function compute_water_storage(csite, lsl,ipa,pool)

And the following lines show how ED2 tracks the water budget. You can search how each term was calculated to help you track water budget based on output.

ED2/ED/src/utils/budget_utils.f90

Lines 895 to 899 in bae4504

	wcurr_residual = wbudget_deltastorage &
	- ( wcurr_precipgain - wcurr_loss2atm &
	- wcurr_loss2drainage - wcurr_loss2runoff &
	+ wcurr_denseffect + wcurr_wcapeffect &
	+ wcurr_zcaneffect + wcurr_pheneffect )

[Manuela-RT]

@xiangtaoxu Thank you for your suggestion, I will check in detail the budget_utils.f90 file.

At some point I actually noticed the variable loss2atm, and I had a couple of questions about how to use it in my case. Maybe you can help me to understand it:

In the ed_state_vars.f90, it is described as "Mean moisture transfer from the canopy air to the atmosphere", but then I was wondering how to deal with ET, canopy air space storage, and loss2atm in a way that I don't end up "double counting" something. When there is water vapor coming from the soil or plant surfaces, does not it first end up in canopy air space and then potentially leaves the patch via loss2atm?

I guess it will all depend on how I define the physical boundary for my water balance. For example, let's say I want my bottom boundary to be the bottom of the deeper soil layer and the upper boundary to be soil and plant surfaces, in that case I guess I should only use ET. But: If I define my upper boundary as the top of the canopy air space, then I should only account for loss2atm.

Am I misunderstanding the loss2atm variable?

Thank you!

[mpaiao]

The budget variables are not part of the mean monthly output, because they are used to track conservation at each time step. In any case, wbudget_loss2atm is numerically equivalent to (−1 × MMEAN_VAPOR_AC).

@Manuela-RT you are correct, the budget will depend on where you define the boundary, so I agree 100% with you. And thinking a bit more on it, I think what you suggested (i.e., exclude canopy air space from the budget) is likely a better approach, because we lack a variable to track it accurately.

The other thing is that you may never get full closure by only looking at the monthly averages, because that would incur substantial double counting. Strictly speaking, you would need to calculate the change of storage as the change in instantaneous values between the beginning and the end of your averaging window. I normally suggest avoiding using the history files for analyses, but this may be needed in your case.

If you save history files every month (i.e., set NL%UNITSTATE = 2 and NL%FRQSTATE = 1.), this is how you may be able to obtain the budget closure:

Let's say for May 2025, this is what you need:

1. History file on 2025-05-01-000000
2. History file on 2025-06-01-000000
3. Monthly mean file for 2025-05-00-000000

Take the monthly average for May 2025
MONTH_DURATION = 31× 86400
MMEAN_INPUT_PY = MONTH_DURATION × MMEAN_PCPG_PY
MMEAN_ET_PY = MONTH_DURATION × ( MMEAN_VAPOR_LC_PY + MMEAN_TRANSP_PY + MMEAN_VAPOR_GC_PY)
MMEAN_OUTPUT_PY = MONTH_DURATION × ( MMEAN_RUNOFF + MMEAN_DRAINAGE + MMEAN_ET_PY )

Using the history file for 1 May 2025
INITIAL_STORAGE = WATER_DENSITY × SUM_SITE( SUM_PATCH( SOIL_WATER_PA(1:NZG,IPA) × AREA_PA(IPA) × AREA_SI(ISI) ) ) + SUM_SITE( SUM_PATCH( SFCWATER_MASS(1:NZS,IPA) × AREA_PA(IPA) × AREA_SI(ISI) ) )

Using the history file for 1 June 2025, find the final storage the same way as the initial storage.

Then the budget should be:

FINAL_BUDGET - INITIAL_BUDGET = MMEAN_INPUT_PY - MMEAN_OUTPUT_PY

A few final considerations:

1. If you are using plant hydraulics, then storage must account for varying canopy water storage too, @xiangtaoxu may be able to help figure out the math for it.
2. Make sure to use the actual month duration for each month, because ED2 uses proleptic Gregorian calendar.
3. You are likely the first person trying to do this, so let us know what you find. Also it's been a while since I thought and worked on these things, I am likely forgetting something...

[Manuela-RT]

@xiangtaoxu @mpaiao Thank you both for taking the time to help me with these detailed suggestions. I marked this as the answer to my question because I think I have more than enough information for what I would like to do. Once I finish the first complete draft of the chapter, I'll give you an update about the results by email.