can project_to_ppi() project to a raster in polar coordinates

[adokter]

request by @fyiguo - maybe terra supports CRSs in polar coordinates?

[bart1]

How about using a polar coordinate system and putting the origin at the radar? With Mercator you can forexample change the origin, something like this:

+proj=merc +lon_0=-180 +lat_0=90 +x_0=0 +y_0=0 +datum=WGS84 +no_defs

What exactly is the goal?

[adokter]

Hi @bart1, for some take-off studies you only use the lowest sweep, and in those situations it can be handy to just keep the exact same coordinate system as the lowest sweep. User is now forced to project on a cartesian grid.

I don't think there are any WKT or proj strings that define a true (range, azimuth) polar coordinate system. So I think instead of a integrate_to_ppi() function we could make an integrate_to_scan() version that outputs a scan instead of ppi. Let me know if you have other ideas.

btw we now use +proj=aeqd +lon_0=radar_lon +lat_0=radar_lat + units=m as the default starting projection.

[bart1]

@adokter One option that I do in some software is keeping/caching one transformation mapping. Here is a small example how I roughly do this. First i create a index scan that only keeps projection relevant attributes (note these are some snippets of code but not a reprex):

      index_scan <- scan |>
        bioRad::calculate_param(INDEX = (1:prod(dim(DBZH)))) |>
        dplyr::select(INDEX)
      index_scan$attributes$what <- NULL
      index_scan$attributes$how <- NULL
      index_scan$attributes$where$a1gate <- NULL
      index_scan$datetime <- NULL
      attr(index_scan$params$INDEX, "datetime") <- NULL
      index_scan

Then I calculate a transformation matrix:

transform_matrix<-bioRad::project_as_ppi(index_scan, grd)$data

This is a expensive operation but can be cached, one can hash the combination of index_can and the target (grd). These should be the same for different timestamps of the same radar. Then future transformation become a simple lookup, something like this:

matrix(scan$params[["DBZH"]][c(transform_matrix)], nrow = nrow(transform_matrix))

This would help for both project_to_ppi and integrate_to_ppi. It would be easiest to then take a dependency on simple caching package like cachem

[bart1]

Something like this could be implemented in scan_to_spdf and scan_to_raster