Next Steps

[Erin-Morrison]

• [x] Explore data.
• [x] Why are there 2000 rows? Should there be another filter condition?
• [x] Improve basic latitude by longitude graph (continent outlines, circumpolar view).
• [x] What is the MAT column? Find and plot its relationship with soil respiration.

[bpbond]

Feel free to tag me with questions @Erin-Morrison and/or check off items above. Thanks!

[bpbond]

Adding first spatial datasets!

• [x] Take the spatialdata WorldClim example and calculate MAT_wc and MAP_wc for your >50N points
• [x] Sanity check: compare these variables with the SRDB's MAT and MAP
• [x] Sanity check: polar plots of MAT and MAP

Linear regression

• [x] Take a look at first few sections of https://rpubs.com/bpbond/830637 code is here https://github.com/bpbond/basicstats
• [x] Our goal would then be to fit a linear model with Rs as a function of MAT and MAP

[Erin-Morrison]

@bpbond I am trying to get the WorldClim data but I keep getting the message "The geodata server seems to be off-line" I even get that message when I try to run the example in the documentation help page.

[Erin-Morrison]

• [x] What are the dominant predictors of your best models? In particular, what’s driving the R2=0.71 (very high!) model?

For this, use car::Anova(mod, type = "III")

• [x] Make some observed versus predicted plots for the models
• [x] Did you try the high-resolution MODIS NPP map? Did it have any better correspondence with SRDB NPP?
• [x] Code and repo cleanup, if needed/useful

[bpbond]

What about making a summary table of p-values from car::Anova(m, type = "III"):

Predictor	Local-NPP model	Satellite NPP model
MAT	< 0.01	<0.01
MAP	n.s.	n.s.
Permafrost	n.s.	...
Soil drainage	0.02	n.s.

[bpbond]

Two or three digit p-values down to 0.001, and below that "<0.001"