davidorme / formind-pfts

Predominantly R code duplicating the FORMIND parameterisation of plant functional types.

formind-pfts

This repo contains predominantly R code duplicating the FORMIND parameterisation of plant functional types. The main point is to provide quick tools to visualise a PFT with given parameters, including the geometry of the PFT and the growth of the PFT under simplified conditions (constant illumination, no water or temperature constraint).

The R code to generate the PFT geometry and growth is in FORMIND_tree_models.R. There are three example folders attempting to match the implementation in R to FORMIND outputs and exploring the growth behaviour.

Tropical Forest

This uses a modified version of the tropical forest parameterisation provided with FORMIND. It has been switched to use the simpler Chanter parameterisation for diameter increments. Additionally, all mortality and seedlings have been turned off and the .pin file has been edited to start with a single sapling of each of the three PFTs in widely separated patches. This gives a single individual growing without shading or competition and makes it easy to study the growth pattern by looking at the contents of the .res file for the simulation.

The R file tropicalForest_noseeds_chanter.R goes through a process of:

1. Matching the geometry of the PFTs from the .res file to the functions in R.
2. Checking the growth mechanism from the FORMIND .res file (that $$D_{n+1} = D_{n} + DInc{n+1}$$ etc.)
3. Simulating the growth using R and comparing that to the .res file. The PDF in the folder shows the differences between the variables calculated in FORMIND and through R are very similar and that the productivity curves through time using the two methods are pretty much in agreement.

There are some unfixed issues - a small but systematic error in the PB calculation and the Chanter model for the intermediate PFT doesn't seem to behave in the same way, although the parameters are the same.

Lonesome

I set this up to test a single tree and then got some odd behaviour in the .res file, which I've been trying to understand. The R file lonesome.R goes through the comparison of the R code to the .res data. The key points are:

1. The production curves from the .res file showed zero maintenance respiration for the first 25 years of growth but growth continues. This then spikes up and growth respiration and growth drop (Top left panel in the Lonesome_Growth.pdf).

   However,R_Growth, R_Main and BInc do still sum to give PB. That's what I'd expect - PB is the budget to be split between growth, growth respiration and maintenance respiration.

2. The basic issue is obvious: the growth parameterisation is wrong so that growth exceeds the available production. If I naively simulate it in R (top right panel in the PDF), negative maintenance respiration is needed to make the budget sum up.

3. I looked at the relationship between D and DInc, in the .res file to see how growth was proceeding and there is a clear truncation of the expected polynomial shape until the tree gets $D=0.6$ at about 25 years.

4. I can reproduce this behaviour pretty closely in R by finding a DInc value that gives growth costs (BInc + R_Growth) that equal the available PB, leaving zero maintenance respiration. If I run this constrained simulation then the production curves match what I see in FORMIND and the simualted DInc also matches closely to the values reported in the .res file.

Tropical Forest overgrowth

Without trying to do any simulation in R, I've also created a version of the tropical forest with a hugely overclocked DInc function. The setup is identical to tropicalForest_noseeds_chanter.par except:

array	N_Par.Pro_dbh_growth
	\u -
	\r -999999:999999
	\d Parameters of the potential growth function of stem diameter per year (group-specific)
	\i 0
	typeOfArray	float
	dimension	4	3
data
	0.42	0.074	0.0215
	2.1	1.5	1.1
	0	0	0
	0	0	0
end

is replaced with:

array	N_Par.Pro_dbh_growth
	\u -
	\r -999999:999999
	\d Parameters of the potential growth function of stem diameter per year (group-specific)
	\i 0
	typeOfArray	float
	dimension	4	3
data
	1.0	0.4	0.085
	2.1	1.5	1.1
	0	0	0
	0	0	0
end

So, the three initial saplings (one of each PFT) in the .pin file are now trying to grow ridiculously quickly, but with no more photosynthetic biomass available as the geometry and illumination are unchanged. Looking the production in the .res file, all three now show the same behaviour as Lonesome:

1. Truncation of the achieved DInc curves from the model - that's expected as the trees simply can't grow this much, but I'd expect them to be even more truncated, to allow for maintenance respiration.
2. All achieve Dmax much earlier. This could make sense if there was 'spare' PB, that growth is now allowed to exploit.
3. All show the same production behaviour - maintenance respiration is zero until growth gets to the point where DInc is achievable within the PB budget.