scripts for running point process models for many species
Citation:
-
Download & clean GBIF data: https://github.com/payalbal/gbifprocessing
-
Download covariate data: data_downloads
-
Covariate data processing: data_processing
-
Model fitting: ppm_fitting
-
Model predictions: ppm_predictions
-
Mapping:
Functions:
Ancillary script files
Notes: 4.Move species occurrence points falling off the mask to nearest 'land' cells We lose data again i.e. number of unique locations is reduced. This can be problematic for ppms...
nrow(unique(outside_pts)) nrow(unique(land))
-
Extract covariates for presence points For landuse: Take the raster value with lowest distance to point AND non-NA value in the raster This takes very logn to run for global layers; find alternative...
-
Catch errors in models Think of how to reduce these. Possible quasiseperation issues due to spatially restricted...could resolved when biuilding models on reduced area? At the moment, the script finds species with errors and reruns the models. If errors persist, species with errors are discarded.
-
LOCATE ERRORS AND RERUN ANALYSIS FOR SPECIES WITH ERROR To be automated if error problem is not solved by species grouping At the moment, it appears that error might be when species data is spatially restricted.