search

EcoClimLab / ForestGEO-tree-rings

Repository for analysis of tree-ring data from 10 globally distributed forests (Anderson-Teixeira et al., in press, Global Change Biology)
2 stars 2 forks source link

Reviewing climwin variable selection #28

Closed teixeirak closed 4 years ago

teixeirak commented 4 years ago

Goal- review the results here to ensure that everything looks reasonable. Also, compare with GAMM output.

teixeirak commented 4 years ago

SCBI

GAMM: image

June- July precip response - makes sense - positive response in both climwin and latest GAMM, consistent with Helcoski. image

May-July cloud response - makes sense - positive response in both climwin and latest GAMM (for most species), consistent with Helcoski. image

May-Aug PET response - probably makes sense -decline at high PET in both climwin and latest GAMM (for most species), consistent with Helcoski. image

bottom line-- I think the SCBI responses all make sense.

teixeirak commented 4 years ago

BCI

latest GAMM: image

July-Aug PET response - similar in climwin and latest GAMM. This is wet season, but should be the hottest month, so possibly temperature stress? image

May WET response - May is beginning of wet season, so potentially makes sense. image

With both, the climate variable distributions look pretty unusual. We should check the CRU data (I'll open a new issue.)

teixeirak commented 4 years ago

Harvard Forest

latest GAMM results (we think we have the structure right, no DBH for now): image vertical bars show mean +/- 1 SD for climate variable.

response to previous June moisture- not sure if this makes sense - It's a unimodal response in climwin and for most abundant species (QURU) in latest GAMM. I'm surprised that this comes out. It's also strange how there's not a patch of time windows with similar responses surrounding it, as tends to be the case at other sites.* image

March to April cloud response - hmmm.... - that would be late spring, maybe affecting spring phenology?? QURU dominates numerically and matches climwin, TSCA apparently likes cloudy springs?? Again, it's also strange how there's not a patch of time windows with similar responses surrounding it, as tends to be the case at other sites. image

March T_max response - hmmm.... - that would be late spring, maybe affecting spring phenology?? QURU dominates numerically and matches climwin, TSCA likes warm March (believable). Again, it's also strange how there's not a patch of time windows with similar responses surrounding it, as tends to be the case at other sites. image

bottom line--I don't see any major problems/ inconsistencies, but the relationships/ variables identified surprise me. I'd like to get input from Neil and @crollinson as to whether these make sense.

teixeirak commented 4 years ago

from Neil, regarding Harvard Forest:

I’ll briefly say that the HF is the most confounding forest in terms of climatic response that I’ve sampled, except for maybe the white oak sitting on thin soils on the top of a low mtn in western North Carolina - no drought signal!

The climate signal at the HF, via traditional tree ring measures, it paltry at best and seems to be going away. During my dissertation, the most recent climate data we could get ended in 1994. The climate signal in red maple at that time was moisture early in the growing season and then a bit of a positive temperature signal during the depths of winter. When we updated the collection in 2014 and did the analysis up to 2012 or so, most of that signal was gone. Compared to the 19 other red oak collections I have around NYS and central and western New England, the HF red oak are the least sensitive to climate. I had an excellent REU who did some experimental climate response analysis. HARDLY any signal in a few species. Something weird here I’ve not put my hands on yet.

teixeirak commented 4 years ago

Lilly Dickey

latest GAMM: image

June precip - makes sense, matches GAMM image

June PET - makes sense, matches GAMM image

bottom line- These results make perfect sense, matching the GAMM and Justin's paper.

teixeirak commented 4 years ago

@ValentineHerr , my conclusion is that this seems to make sense overall and be working as we want. I still want to hear what some of our dendro collaborators have to say about it, but I think we can move forward on the assumption that we've got it about right.

@biancaglez, reading through my posts above would be a good way for you to get an overview of what we're finding so far.

teixeirak commented 4 years ago

I should note, though, that Scotty Creek is the site where I'm most concerned about climate variable selection (see issue #25).

teixeirak commented 4 years ago

Next round of review: issue #38.