PecanProject
Migrate geospatial tools from the rgdal ecosystem (rgdal, rgeos, maptools) depends to sf & terra
Description
Apparently, the R spatial world is shifting away from rgdal, rgeos, and maptools toward sf and terra. rgdal, rgeos, and maptools all scheduled for retirement at the end of 2023. Thanks for the heads up @infotroph
https://mobile.twitter.com/RogerBivand/status/1407705212538822656 https://cran.r-project.org/web/packages/rgdal/index.html
raster doesnt yet have a retirement date, but its suggested to move toward terra
Proposed Solution
So to me this means right now is a great time to overhaul our spatial dependencies. First start with simpler functions toward sf/terra (e.g. in data.remote) but we need to develop a list of all functions that use these tools and then schedule the order for updating.
Alternatives Considered
none
Additional Context
[](https://twitter.com/@RogerBivand)[Roger Bivand](https://twitter.com/@RogerBivand)Roger Bivand [@RogerBivand](https://twitter.com/@RogerBivand)
[@edzerpebesma](https://twitter.com/edzerpebesma) I'm not sure that taking over maintenance is a sensible use of effort (add maptools to that list). The main condition for this will be transition from raster to terra, as raster uses rgdal heavily. If sf/stars and terra become the main workflow bases, things get simplified.
TwitterTwitter | [Jun 23rd, 2021](https://mobile.twitter.com/RogerBivand/status/1407705212538822656)
Some other issues - if we need to buffer points to average pixels when extracting does terra support that like raster?
Appearances of rgdal:
https://github.com/PecanProject/pecan/blob/1043bfd57becbb79885e9455439308c68b9816ca/modules/data.land/R/gis.functions.R#L71-L72
https://github.com/PecanProject/pecan/blob/1043bfd57becbb79885e9455439308c68b9816ca/modules/data.remote/R/NLCD.R#L72
https://github.com/PecanProject/pecan/blob/1043bfd57becbb79885e9455439308c68b9816ca/modules/data.atmosphere/R/GEFS_helper_functions.R#L248
Plus a LOT of scripts and vignettes
It's October. rgdal will disappear from CRAN soon and CI will start breaking.
@DongchenZ @Shuohanz I'm assigning the GEFS part of this to you two, since you're actively working on that code already.
The NLCD and shp2kml bits are not core so could be temporarily deactiviated if no one steps up.
