Meeting notes copy

[jianqiuz]

===May 1st===

General discussion on work scope, interesting directions to follow up:

1. Representation of reaction kinetics in microbial models of SOM decomposition
2. Development of continuous carbon pools in SOM models
3. Application of probability/ distribution functions to represent continuous features in SOM models
4. Machine learning applications for global scale modeling

[kdorheim]

May 15

Status/Questions

• Made my way through Georgious 2017 and Shi 2018 paper have skimmed the other papers but wanted to familiarize myself with the MEND model first Wang et al 2013
  • Question on the Shi 2018 paper figure 4 what and the uncertainty ranges
• Sketching out and thinking about the MEND model design
  • What pieces of the model set up should be flexible? The equations that modify the fluxes Equations 9 - 17?
  • Are there other models you have worked with in R in the form of an R package? If not that is okay but that might be useful to look at

Meeting Notes

• There are lots of different data sources that we could use could collaborate with BBL and Will Wieter, also might be interested in looking at nutrient addition data sets
• May incorporate N and P dynamics into the model
• Will start by developing the model published by the Wang 2013 paper and reproducing some of the simple simulations
  • will want the ability to change the representation of the fluxes (F)
    • eq 9 may change to ECA
      • eq 14/15 for desorption/absorption could have absorption completing with microbial update
      • Eq 12/13 use CUE we might want to include some representation of how environmental conditions impact CUE

@jianqiuz please feel to edit and add if you feel I've left something out.

[kdorheim]

May 26

I've made progress on developing the model as an R package, there is one more model configuration I would like to try before doing a more detailed set of validation runs between our our results and Wand et al. 2013 results.

[kdorheim]

FWI this was the first attempt

[kdorheim]

June 12

• Established that the order of the input variables was causing the funky results.
• Next Steps
  • Set up ability to optimize MEND to some data (ideally use the mcmc package)
    • nonlinear
    • interference analysis mcmc
  • Flexibility
    • MIMICS kinetics / reverse MM kinetics
    • Modify the number of carbon pools
• Thinking about pubs
  • GMS or ESSD

[kdorheim]

July 10

Ideas

• Develop and release the MENDplus package, a tool that lets users examine the impacts of structural and parametric uncertainty on soil carbon dynamics.
  • Submit a documentation paper? We discussed GMD but other ideas include PLOS ONE, JORS, JOSS, or ESSD.
  • A write up demonstrating the different simulations and optimization capabilities results.
• Use MENDplus to obtain a global estimate?
• Can we use MENDplus to help try and figure out the partition of biomass?

Logistic Questions

• PNNL software approval?
• Collaboration with the original authors of MEND? What do we do to indicate the legacy of the model but also make sure that it is different enough? Check in with BBL and Will?
• Beta testers?

Action Items

• Set up DOM with R-MM dynamics
  • Compare the default and DOM R-MM over a series of the ratio of DOM to BM. Determine if there is a tipping point?
• Set up the MCMC optimization first using CO2 data eventually we may want to use BM observations but those observations may be limited

@jianqiuz Please feel free to add/expand 👍

[kdorheim]

July 24

Agenda & Status Update

• Check in about ~ BBL meetings?~ ~ LDRD status?~ ~ Summer school?~ ~ DOM with the R-MM dynamics see (https://github.com/kdorheim/MENDplus/issues/9)~ ~* Started setting up the MCMC, right now stuck on playing around with the step/batch size to get the acceptance rate right. What is going to be helpful now it probably determining what priors we want to be using...~

Recap and Action Items

• Talked about JZ's summer school, which gave us some ideas about adding a visualization platform may be using Shinny. @jianqiuz do you have the link to the youtube videos? It would be cool to check some of them out, I am most interested in the DOM video you mentioned.
• MCMC
  • [ ] KD to add some user documentation and push to github
  • [ ] JZ play around with and give KD some feedback
  • [ ] Identify the parameters to calibrate and assume priors (right now we are thinking BM related params)
• Change pool structure
  • Is there a way to easily implement in code? May be we can add DOM sub model components... with only an aggregate DOM flux affecting other pools...
• Set up meeting with BBL

[jianqiuz]

Aug 21 Database links: Soil Incubation Database https://github.com/SoilBGC-Datashare/sidb Global soil respiration flux data https://daac.ornl.gov/SOILS/guides/SRDB_V4.html EMSL summer school https://www.youtube.com/user/EMSLatPNNL/videos