Changes to N-fixing penalty function and salinity tolerance

[MichaelBarryBMT]

Hi Matt and Casper

A couple of requests on phyto routines in this one:

• Matt and I discussed the metabolic penalty function for nfixers. Although I understand the double engine effect, I think the requested form of the function is what Matt suggested it should be when we chatted
• The logic salTol >=4 means that all except one fSal is applied to respiration. This does not seem to match up with the shapes of the fSal functions. If this was the case, then most of these functions will have unlimited growth, and respiration occurring only in small salinity bands - which without limitation by nutrients/temperature etc can lead to large phyto growth. The suggested change may also not be correct - but review of the logic is appreciated please

Both of these are just suggestions - I just need to know the reasoning behind whatever is correct so that I can finish writing the WQM manual and close this issue out please.

Ta

MB

[matthipsey]

Hi @MichaelBarryBMT

We've done a deep dive into this and made a few changes to clean it up.

In short, the form of the expression is sensitive to S_bep. If S_bep <1 then 0 < f(S) < 1, but if S_bep > 1 then f(S) >1 (except for the salModel 4 which is different).

What Ive done is made a new rule for this :

• If salTol is -ve, then apply the function as growth suppression : f(Sal) <= 1 will reduce GPP
• If salTol is +ve then apply the function as respiration enhancement : f(Sal) >= 1 will increase RSP

The user needs to be careful to set S_bep accordingly (either less than or greater than one) or they will get the opposite effect - there is a warning, but this may get missed.

salmodel 4 is a special case, but we've also made a way so that it conforms to the above logic.

I''m not sure how choosing the growth suppression or resp enhancement will affect the calibration process and will need to do some testing to get an idea.

Maybe for your implementation then you could hide one and just rely on a particular form.

https://github.com/AquaticEcoDynamics/libaed-water/commit/712481383d18858414a874cc7ad837fd6b686d7c

[aed-modeller]

A few further notes on the top of Matt's response:

1. The option "salTol == 1" only works for freshwater, e.g. salinity < S_maxsp;
2. The option "salTol == 2" or "salTol == 3" work for estuarine water and adapt to a wide range of salinity;
3. The option "salTol == 4" only works when S_bep < 1

The S_bep is a key parameter for the fSal, when S_bep > 1 then it acts as an amplifier, otherwise is a suppresser.

Below is an example graph showing the response of fSal to S_bep in all salinity limitation functions (note that in future we will use "salTol == -1/-2/-3" for respiration enhancement, and "salTol == 1/2/3/4" for growth suppression):

[aed-modeller]

I think we need to add a few lines in the "salTol == 1" option to avoid the fsal goes down below 0. E.g.

IF (salinity>phytos(group)%S_opt .and. salinity<phytos(group)%S_maxsp) THEN
        tmp1 = (phytos(group)%S_bep-1.0) / ((phytos(group)%S_maxsp - phytos(group)%S_opt)**2.0)
        tmp2 = (phytos(group)%S_bep-1.0) * 2.0*phytos(group)%S_opt / &
              ((phytos(group)%S_maxsp-phytos(group)%S_opt)**2.0)
        tmp3 = (phytos(group)%S_bep-1.0) * phytos(group)%S_opt*phytos(group)%S_opt / &
              ((phytos(group)%S_maxsp-phytos(group)%S_opt)**2.0) + 1.0
        fSal = tmp1*(salinity**2.0) - tmp2*salinity + tmp3
     ELSEIF (salinity>=phytos(group)%S_maxsp) THEN
        fSal = 0
     ELSE
        fSal = 1.0
ENDIF

[matthipsey]

Hi @MichaelBarryBMT

Ive had a close look at this.

I think it is implemented correctly in the code as it was, and your suggested modification was not right...

k_nfix is the fractional rate of GPP under full N-fixation (eg 0.8 is 80% efficient growth rate when fixing)

GPP = GPP * ( k_nfix + f(N) [1-k_nfix] )

when f(N) is 1, then the amount in the ( ) sums to 1, and no penalty is applied (as no n-fixing needed)

when f(N) is 0, then the amount in the ( ) sums to k_nfix, and the penalty is fully applied

[MichaelBarryBMT]

Hi Matt Thanks for looking at this. Agreed! MB