output mean canopy and understory mortality rates

[ckoven]

on the same note as #128, we ought to output the mean mortality rates separately for understory and canopy to compare with obs. possibly some other vital rates too. right now we have these by DBH class, but a simple canopy-level mapping would also be useful.

[tompowell9]

Perhaps this is already an output, but can we have the mortality output at the cohort level? This way the end user can decide how to assign the bins, both in terms of size and PFT, for comparing to site-specific observations. For example, Condit et al. 1995 reports mort_rates for 1-10cm and >10cm trees for different PFTs at BCI; but Nepstad et al. 2007 reports mort_rates in 2-5 cm, 5-10 cm, 10-30cm and >30cm bins, but does not separate into PFTs. With cohort level mort_rates the end user can compare directly to each site by binning accordingly in their own post processing scripts. -TP

On Wed, Oct 5, 2016 at 2:39 PM, Charlie Koven notifications@github.com wrote:

on the same note as #128 https://github.com/NGEET/ed-clm/issues/128, we ought to output the mean mortality rates separately for understory and canopy to compare with obs. possibly some other vital rates too. right now we have these by DBH class, but a simple canopy-level mapping would also be useful.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/NGEET/ed-clm/issues/130, or mute the thread https://github.com/notifications/unsubscribe-auth/AJnK6AtSgoYZZ7LhAa452qgIw-6BUMLlks5qxBktgaJpZM4KPUaI .

[ckoven]

tom-

outputting anything at the cohort level is currently not possible--since the cohorts aren't persistent, there isn't any way to track them via the history files. instead, we need to bin them somehow for history writes. it ought to be easy enough to change the size class bins to accomplish what you are describing though.

The reason the size-bin approach can't work for canopy vs understory mortality is that what happens now is that, because of PPA, when any closed-canopy cohorts grow, some fraction of the shortest canopy gets demoted, whereupon that newly-demoted cohort experiences elevated carbon-starvation mortality. but it is basically still the same size as its former cohort-mates who are still in the canopy, so would fall into the same size bin as them.

Charlie

[djj4tree]

Charlie in reference to your last point about demotion and carbon starvation. Is that a gradual process or does the cohort go from the lights being on to off? It seems to me it should be a very slow process of being overtopped. Also, are cohorts initiated already in the understory or does that require a disturbance?

[mdietze]

Getting the rate of carbon starvation right is actually very important for 2 reasons: (1) to get the self-thinning slope correct and (2) to get density dependence right. And really (1) is just a special case of (2), but (2) is also important in the lowest parts of the understory -- my experience with ED2 is that it is possible for the rate of recruitment to grow to be greater than the rate of understory mortality due to carbon starvation. Since ED2 has no other implicit or explicit mechanisms for density dependence (e.g. modeling pests & pathogens [see Dietze & Matthes 2014 Ecology Letters] or making germination light dependent), the result is a runaway understory. But you need to balance this need for high mortality in these cases with the need for fairly high resilience in adults. And you can't solve it by just forcing the seedling storage pool's capacity to be tiny, because in reality it's not -- shade-tolerant seedlings do persist in the understory for decades once they establish, often based on very short periods of productivity being shunted to storage.

Admittedly, all of this doesn't solve Charlie's immediate problem with PPA, but just trying to avoid solutions that lead to other, known, errors.

[rosiealice]

Dan - no, in the modified version of the PPA that I made for FATES, at each timestep when the canopy is closed, some area (the new crown growth) needs to be demoted into the understory. In the standard PPA, cohorts -are- demoted all at once when they fall below z - the notional height of the canopy/understorey interface. In FATES, we instead demote a small amount of -every- cohort, but the amount that we demote is a function of cohort height. Taller cohorts have a smaller fraction demoted than smaller ones. There is a parameter that controls the dependance of demotion fraction on height, which in principle is a proxy for the deterministic nature of the system. Small dependencies on height suggest more 'neutral' growth-resource acquisition feedback, and vice-versa.

Mike - I think the phenomenon you note - (recruitment rate> carbon starvation mortality rate) is also quite possible in FATES. In TREEMIG, there is an 'antagonist' term which reduces the seed bank when it is large, and we had some time ago talked about putting a similar thing in FATES. SEIB has a feature which stops the seeds from germinating if there isn't enough physical space for them. I'm sure there are other methods. Getting the understory mortality and the seed germination rates out of the model should at least let us interrogate whether we are in that regime or not...

*until all the trees reach their maximum down area, which is plausible a problem for generating realistic self-thinning curves and we need to address.

On 6 October 2016 at 08:44, Michael Dietze notifications@github.com wrote:

Getting the rate of carbon starvation right is actually very important for 2 reasons: (1) to get the self-thinning slope correct and (2) to get density dependence right. And really (1) is just a special case of (2), but (2) is also important in the lowest parts of the understory -- my experience with ED2 is that it is possible for the rate of recruitment to grow to be greater than the rate of understory mortality due to carbon starvation. Since ED2 has no other implicit or explicit mechanisms for density dependence (e.g. modeling pests & pathogens [see Dietze & Matthes 2014 Ecology Letters] or making germination light dependent), the result is a runaway understory. But you need to balance this need for high mortality in these cases with the need for fairly high resilience in adults. And you can't solve it by just forcing the seedling storage pool's capacity to be tiny, because in reality it's not -- shade-tolerant seedlings do persist in the understory for decades once they establish, often based on very short periods of productivity being shunted to storage.

Admittedly, all of this doesn't solve Charlie's immediate problem with PPA, but just trying to avoid solutions that lead to other, known, errors.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/NGEET/ed-clm/issues/130#issuecomment-251984286, or mute the thread https://github.com/notifications/unsubscribe-auth/AMWsQwrdiDpv004RaMNmHIIyDmRKVYg-ks5qxQk1gaJpZM4KPUaI .

---

Dr Rosie A. Fisher

Terrestrial Sciences Section Climate and Global Dynamics National Center for Atmospheric Research 1850 Table Mesa Drive Boulder, Colorado, 80305 USA. +1 303-497-1706

http://www.cgd.ucar.edu/staff/rfisher/

[rgknox]

@tompowell9: following up on @ckoven 's thoughts. For history (time and space averaged diagnostics, not to be confused with ED2s' "history" which is used for restarting runs) output diagnostics, there are variables that we can generate that bin by size (as @ckoven mentioned), but this binning is also by PFT, a two dimensional matrix. So hopefully this will be helpful if you want to make comparisons between model output and specific trees or tree groups.

Note that there are also ways to generate the model's restart information at higher frequencies as well. The restart files contain an instantaneous snap-shot of the state of the ecosystem, and from these files you can grab cohort-level information on canopy structure and composition. Of course, we should be careful with these types of data, as they are snap-shots and any rates stored in these datasets are only reflective of an instant in time (I think their are mortality rates stored in these files, but I would be very hesitant to use it).

[ckoven]

ditto, this was solved by #184