search

forc-db / GROA

This repository houses data and code for the Global Reforestation Opportunity Assessment (GROA) led by Susan Cook-Patton of the Nature Conservancy.
Creative Commons Attribution 4.0 International
31 stars 10 forks source link

add litter_carbon to variables being incorporated to ForC #7

Closed teixeirak closed 4 years ago

teixeirak commented 5 years ago

@ValentineHerr , @CookPatton,

As Abby will soon be working with the litter/ deadwood data, we'll now want to add those variables. I think it will be best for Abby to reviews the data in ForC format and keeps a list of any corrections that need to go into GROA, rather than trying to master two databases / mess with GROA directly.

@CookPatton:

  1. I noticed that there are a few cases where the covariates "components", XXX were shifted over one column from where they were intended such that "components" goes in covar2_value, XXX goes in covar3.
  2. Another minor cleanup issue: you sometimes use component and sometimes components.
  3. Please review assignments to ForC variables in variable_name_conversion.
  4. If a study reported standing dead wood, how would that be classified?

@ValentineHerr, I've updated variable_name_conversion to include litter components. Could you please revise your code accordingly? Note that we need to look to the covariates to determine which variable is assigned. Also note points 1 & 2 above.

CookPatton commented 5 years ago

@teixeirak @ValentineHerr

Glad to hear Abby is tackling litter/deadwood! I like your proposed approach and can fix any issues Abby spots in GROA.

I'll error check litter/deadwood today and send you the latest data. Also I'll sum soil data to the surface and resend. And fix the covariate issues/component(s). Thanks for spotting those. Standing dead, if described separately would be in dead wood/CWD.

Remind me where variable_name_conversion is - to double check. Thanks!

ValentineHerr commented 5 years ago

@CookPatton, variable_name_conversion is here.

ValentineHerr commented 5 years ago

@teixeirak , Can you confirm that the NA in line 11 of variable_name_conversion file means that the record should be ignored if GROA_variables.name is _aboveground_carbon + littercarbon and component covariate value is either litter, or forest floor, or CWD or FWD + CWD, orFWD+CWD?

thanks

teixeirak commented 5 years ago

That is correct; any NA should be ignored. In this case, we need both litter/forest floor and CWD(+FWD).

CookPatton commented 5 years ago

@teixeirak @ValentineHerr I am about to push the cleaned up litter/deadwood data. I've deleted the sections related to component(s) since that information is now in the variables.name. The variables.name should either say litter_carbon (or litter_biomass) or CWD_carbon (or CWD_biomass), or both.

I also checked the high values for litter and fixed a few errors. There is one paper with 100 tC in the forest floor!

CookPatton commented 5 years ago

@teixeirak @ValentineHerr I should have added that litter versus CWD follows this rule: We parsed data according to IPCC guidelines, where coarse woody debris includes wood lying on the surface, dead roots, and stumps, larger than or equal to 10cm. Litter includes all non-living biomass that is distinguishable from mineral soil, typically 2mm or greater and less than 10cm.

teixeirak commented 5 years ago

Regarding that 100tC in the forest floor... wow! Where is that? Are they including decomposing logs?

CookPatton commented 5 years ago

Boreal Canada - this paper here. They say that they passed the forest floor through a 1.5 cm sieve so presumably just the little bits. What do you think? Can that be at all correct?

teixeirak commented 5 years ago

I'm not sure... That's higher than anything we have so far, but that doesn't mean its wrong. It would be good to do some back of the envelope calculations as to whether that's reasonable given the humus depth and typical bulk density values.

Beyond that, I think we should leave it for now. If it seems to be a big outlier when we look at the broader context, we can reassess/ flag it as suspicious (i.e., exclude from analyses)