Compute initialization after even aged harvest

[kunstler]

=> Georges with FUNDIV data

[kunstler]

I eventually manged to work on that. @jasperfuchs Sorry for the delay.

What I did. I took the fundiv data and for each species selected plot with a mean diameter less than 12cm and at least 90% of the basal area of the selected species. Based on this plot I computed the mean log(DBH) (weighted by the weight of teh trees) and the sd of log(DBH) (with weight as well).

This give me for each species the mean and sd of the size dist (in log(dbh)) of stand with only small tree of teh species.

Then you can simulated a pop by drawing in normal of these param and take exp.

This gives

The red vertical line is the lower size boundary so all trees below this size must be removed.

Then the question is the time it takes to reach this population from the plantation.

I think we can use the recruitment lag of each species and say that with plantation this is shorter. Something like 1/2 the regeneration lag without plantation ? I didn't had the time for look for info in the literature on that point.

@jasperfuchs let me know what do you think of that.

[kunstler]

Of course the pb with this approach is that you will have the population appearing suddenly a certain number after the start of the simulation.

[jasperfuchs]

Thanks, @kunstler!
From my perspective, this is a good approach to derive the initial diameter distribution after clear-cut/planting.

I would opt for applying the species-specific recruitment lag you have already implemented in IPM. This lag should be shorter if the forest is planted. How about using a fixed number of years. We mostly plant trees that are 2 or 3 years old. If your recruitment lag accounts for an additional time lag between seed fall and germination (e.g., 1 year), we could thus reduce the default recruitment lag by 3 or 4 years. The implemented delays seem to be around 15-30 years, so ½ the time may overestimate the growth advantage of planting? What do you think?

Regarding the suddenly appearing population, Carola and I expect that this would have no or at least only a very minor impact on the economic outcome. So, let’s start with this approach and if any strange effects appear in the simulations, we can still use a more complex approach.

[gowachin]

This dataset and a function to exploit it is now present is the package on the dev branch.
An example is here in dev/dev_jasper_b.Rmd. Dataset is called matreex::distrib_planting.

The new function act as an initiation function and can be given when building a species object. This allow to scale the initiation to the plot surface. However, this function will always give the same distribution for a given mesh and plot surface.

Because of this limitation, you can also draw this distribution like illustrated below, and give it to def_init_k() that will initiate a population with a distribution that it takes as an argument.

plant_funct <- def_init_planting("Fagus_sylvatica")
distrib_Fag <- plant_funct(Fagus_IPM$mesh, SurfEch = 1) # 1ha plot is here important !

 Fagus = species(IPM = Fagus_ipm, 
                        init_pop = def_init_k(distrib_Fag),
                        harvest_fun = Even_harv, disturb_fun = disturb_fun,
                        disturb_coef = "sp", rdi_coef = "sp")

This code is a bit more complex, but if you need to scale the distribution for a given number of stems or basal area, it's more convenient. Expected basal area and number of stems is indicated in he dataset with species parameters.

Note that the dataset parameters are computed for monospecific stands.

[gowachin]

Concerning the lag, it's not straightforward to reduce it because it affect the time ellipse after a final cut and the recruitment during growth. One argument could be to deactivate the recruitment and edit the lag, but I don't know if recruitment is totally absent from even managed stands.

Another approach would be to edit the simulations, by reducing the time ellipse afterward. This can be done by removing the number of rows needed and editing the t column.