model2netcdf.ED2() only converts data for first PFT

[Aariq]

Bug Description

I open one of the .h5 files output by ED2 and confirm that there are multiple PFTs listed, then run model2netcdf.ED2() where pfts = settings$pfts, and the resulting .ncvar file only lists the first PFT. The .nc file also only has data for one PFT. I think I tracked this down to this line, which would only read the first pft name.

pecan/models/ed/R/model2netcdf.ED2.R

Line 954 in fbfcd2f

pft_names <- pfts$pft$name

This should be something like pft_names <-pfts %>% purrr::map(~.x[["name"]]) or whatever the base R equivalent is.

To Reproduce

It can be reproduced with just one of the .h5 files output by ED2, but the example I'm using is on a server and I'm not sure how to share it. I put that single h5 file in a dir called "testing" and ran this:

library(ncdf4)
library(PEcAn.ED2)
settings <-
  list(
    run = list(
      site = list(
        lat = 35.9588,
        lon = -84.2874
      ),
      start.date = "2019/03/01",
      end.date = "2019/09/01"),
    pfts = list(
      pft = list(
        name = "SetariaWT",
        ed2_pft_number = "1"
      ),
      pft = list(
        name = "ebifarm.c3grass",
        ed2_pft_number = "5"
      )
    )
  )

before <- nc_open("testing/analysis-E-2019-03-00-000000-g01.h5")
ncvar_get(before, "PFT")
## [1] 1 5
nc_close(before)

model2netcdf.ED2(outdir = "testing",
  settings$run$site$lat,
  settings$run$site$lon,
  settings$run$start.date,
  settings$run$end.date,
  settings$pfts 
)

# 2022-07-20 20:54:11 INFO   [debugscript.R#28: model2netcdf.ED2] : 
#   ----- Processing year: 2019 
# 2022-07-20 20:54:11 INFO   [fcn] : 
#   *** Reading -E- file *** 
# 2022-07-20 20:54:11 INFO   [debugscript.R#28: model2netcdf.ED2] : 
#   *** Writing netCDF file *** 

after <- nc_open("testing/2019.nc")
ncvar_get(after, "PFT")
# [1] 1
nc_close(after)

Expected behavior

Should convert data for all PFTs

[dlebauer]

There are already some sample output files in https://github.com/PecanProject/pecan/tree/develop/models/ed/tests/testthat/data, if those don't meet your needs - e.g. doesn't have multiple pfts, consider adding a (small) new one that does.

[Aariq]

I did eventually find those. I'll check tomorrow to see if one has multiple PFTs

[dlebauer]

Try setting NL%ATTACH_METADATA to 1 for generating test data; otherwise the outputs and their dimensions are difficult to interpret.

[Aariq]

Looks like the current ED output in testthat/data doesn't work at all unfortunately and all tests are currently skipped. Might need help generating new test output.

[Aariq]

Related to #774?

[Aariq]

Won't fix until #1329 is (partially) dealt with.

[Aariq]

So I won't be able to get the tests working because this function is more broken than I can handle without downgrading my R version (see #2981). Should I go ahead and make a PR with the fix to line 954 anyway?