Autocorrelation Plots - Githubissues

cehatch / Soil-Moisture-Paper

Soil moisture research from Foothills Preserve and Tidmarsh Nature Sanctuary, manuscripts

0 stars 0 forks source link

Autocorrelation Plots #23

Open cehatch opened 6 years ago

cehatch commented 6 years ago

[x] Regime separation (chunking), where are ditches? Regimes ( = "Chunks") are separated into DRY - BOG - WOODS
[x] Include behind the barn in Dry or Woods? DRY
[x] AC Plot for: WEST BOG (1 meter)
[x] AC Plot for: EAST BOG (1 meter)
[x] ~~AC Plot for: BOG (10 cm; not whole bog)~~
[x] AC Plot for: WOODS (1 meter)
[x] ~~AC Plot for: DRY (10 cm; only West side)~~
[x] AC Plot for: DRY (1 meter)
[x] AC Plot for: Hummock transect (Tidmarsh East)
[x] AC Plot for: Undisturbed transect (Tidmarsh East)
[x] AC Plot for: AWC transect (Tidmarsh East)

ItoErika commented 6 years ago

NOTE: Can't do cm AC plots because the 10 cm scale was only used every other meter in the West part of the transect.

ItoErika commented 6 years ago

Distances used for regimes in Autocorrelation plots:

West Dry: 0 - 29 m (Luke distance) West Dry is the area where the DTS plow started. West Bog: 46.4 - 143 m (Luke distance) East Bog: 145 - 647 m (Luke distance) Middle Dry: 650 - 701 m (Luke distance) (not to 709 b/c Luke didn't take measurements on road) Middle dry is the dry area between the woods and the bog Woods: 709 - 777 m (Luke distance)

ItoErika commented 6 years ago

TW AC Plots: TW_DryW_AC_m.pdf TW_BogW_AC_m.pdf TW_BogE_AC_m.pdf TW_DryM_AC_m.pdf TW_Woods_AC_m.pdf

cehatch commented 6 years ago

Tidmarsh_East_Transects_GSM.xlsx

ItoErika commented 6 years ago

TE AC Plots: TE_Hummock_20cm_AC.pdf TE_Undisturbed_20cm_AC.pdf TE_AWC_20cm_AC.pdf

ItoErika commented 6 years ago

Done! @cehatch

cehatch commented 6 years ago

These look great, @ItoErika ! Thank you!!! First thoughts on interpretation: 1) The "typical" walking mean shape where there is a monotonic decrease in value means that points are increasingly UNLIKELY to predict each other with increasing distance either in the correlated (positive) or anti-correlated (negative) sense. 2) The relatively flat shapes mean that each point is almost as likely as each other point to predict the next one, although no particular distance-dependent trend is present (e.g. they all are pretty close to the mean value, and don't vary very much). 3) The 95% confidence bounds are larger (less precise) if there are few data points, and tighter/narrower (more precise) if there are more. I'm not sure exactly how these bounds relate to what's on the plot - should they be the "error bars" on each dot?