Time series model in R by site

[swood-ecology]

Run time series model in R at site level for four test sites.

[eoway]

To clarify, this would look at changes in peak productivity trends over time as a function of SOC+ clay fraction + drought + RMZ (random effect). The goal is to conduct an analysis comparable to what's be conducted in the midwest row crop system.

[swood-ecology]

Is the unit of observation the grid cell, with the site as some kind of geographical control? There might not be much credible variation in soil properties at that fine of a scale. Put another way, the variation in soil properties is likely going to be more dependent on veg indices at finer scales because there are fewer actual soil samples at those scales to inform the interpolation. Worth trying out, but this kind of model might be more credible at the whole-state level.

[eoway]

Ok, but I thought we were interested in whether some areas have more or less variation in annual peak productivity over time, with a goal being to identify whether SOC and/or substrate (clay fraction) was a driver of that variation. Is that still what you're thinking, but looking at variation at the statewide scale?