Annual biomass not included in previous year's biomass passed to SOILWAT2 each year

[kpalmqui]

In ST_main.c, annual species biomass is accumulated via stat_Collect before annuals are killed each year, such that annual species biomass is non-zero in the bmass output files. Thereafter, annual species are killed via _kill_annuals.

THE PROBLEM: the biomass values that are passed to SOILWAT2 for annuals are the end of year values after annuals are killed (i.e. 0) and reduce biomass in SOILWAT2 compared to what was present that year in STEPPE.

Essentially, we need to modify the code such that the biomass values for annual species going into SOILWAT2 each year (i.e Env_Generate) represent establishment this year (already in place) and last year’s biomass. The good news is that we do save annual species relsize to calculate last year’s biomass via save_annual_species_relsize for use in annual establishment, so it shouldn’t be too hard to implement.

I will post a more formal solution shortly.

[kpalmqui]

Solution:

In sxw.c (SXW_Run_SOILWAT) include last year's biomass in sizes[g] by replacing:

#this correctly reflects perennial biomass and represents annual biomass due to this year's establishment only
ForEachGroup(g) {
        sizes[g] = RGroup_GetBiomass(g);
}

with:

#for annuals, get this year's biomass due to establishment
ForEachGroup(g) {
        sizes[g] = RGroup_GetBiomass(g);

#if the species is an annual, augment sizes[g] with last year's biomass        
        ForEachEstSpp(sp, g, j) {
            
            if (Species[sp]->max_age == 1) {

                sizes[g] += Species[sp]->lastyear_relsize * Species[sp]->mature_biomass;
            }
        }
    }

Using print statements I am seeing increments in ~1g per annual species per year (~4g) with the default inputs.

@dschlaep let me know if you see problems with this implementation.

[dschlaep]

@kpalmqui - I agree that your solution and commit 21ac106 makes that annuals and perennials are finally treated equivalently in terms of biomass passed to SOILWAT2.

Biomass passed to SOILWAT2, as your code comments in the commit explain, represent now last year's biomass plus biomass due to current year's establishment -- that is because Env_Generate() is called at the beginning of a year, but after rgroup_Establish().

If most biomass is perennial, i.e., current year's establishment adds little biomass relatively, then this is ok; however, if current year's establishment adds biomass that is a substantial fraction of last year's biomass (let's say x% >> 0), as is the case for annuals, then we now pass on average about (100 + x)% biomass to SOILWAT2. As long as we simulate sites where annuals represent a smallish proportion of the vegetation, then this overestimation should be unproblematic. However, I don't currently see a better way than how you fixed it.

[kpalmqui]

@dschlaep thanks for taking a look and for the comment! The only other possibility would be to move Env_Generate() to before rgroup_Establish(), which would cause new and more significant problems I think.

I'll open a pull request and tag you as a reviewer. Then I will merge into master and then feature_space etc.

[kpalmqui]

closed by 21ac106