Open emmamendelsohn opened 1 year ago
Deciding not to use grazing capacity as it is not informative - it predicts livestock coverage, which is redundant with GLW. We also discussed correcting the model for livestock population, i.e., being able to predict which locations would have an outbreak if there were livestock. This is more about informing decisions for where to have livestock rather than providing an alert for an immediate outbreak.
slope and aspect: https://www.fao.org/soils-portal/data-hub/soil-maps-and-databases/harmonized-world-soil-database-v12/en/
key to soil data, using drainage(p12) and texture class (p15) : https://www.fao.org/3/cc3823en/cc3823en.pdf
We should verify the soil type key and aggregation strategy. Some of the groupings conflict (5 -> 1 but then later 10 -> 5 and 7->2 but later 12->7). The key is pulled from the key above. @noamross @emmamendelsohn
# At this point:
# 1 - clay (heavy)
# 2 - silty clay
# 3 - clay
# 4 - silty clay loam
# 5 - clay loam
# 6 - silt
# 7 - silt loam
# 8 - sandy clay
# 9 - loam
# 10 - sandy clay loam
# 11 - sandy loam
# 12 - loamy sand
# 13 - sand
# Re-code factor levels to collapse simplex.
# Figure out where key is for the units are in HWSD2
dat_out$HWSD2 <- if_else(dat_out$HWSD2=="5", "1", # clay (heavy) + clay loam
if_else(dat_out$HWSD2=="7", "2", # silty clay + silty loam aka
if_else(dat_out$HWSD2=="8", "3", # clay + sandy clay
if_else(dat_out$HWSD2=="9", "4", # silty clay loam
if_else(dat_out$HWSD2=="10", "5", # clay loam + sandy clay loam BUT SEE RULE 1!!!
if_else(dat_out$HWSD2=="11", "6", # silt sandy + loam
if_else(dat_out$HWSD2=="12", "7","0"))))))) # loamy sand + silt loam BUT SEE RULE 2!!!
We may have decided not to include bioclim and forest cover because its captured by NDVI. Does that ring a bell @noamross ?
I'm trying to remember. But given that both climate and forest info are encoded in the dynamic variables, that makes sense. I'd put off putting those two in.
Slope and aspect will not be relevant/interpretable at the ADM level. If they are already working and available, great, but I wouldn't spend any time troubleshooting them.
For each, find data source, download and normalize to project spatial resolution (0.1 x 0.1)
Then, PCA, and visualize results, interpretation. We will decide number to use variables.