write n2o and ch4 flux variables to SIPNET netCDF#3816
Conversation
| outdir <- withr::local_tempdir(pattern = "sipnet_out_") | ||
| rundir <- withr::local_tempdir(pattern = "sipnet_run_") | ||
|
|
||
| # minimal sipnet.param — model2netcdf only reads leafCSpWt for LAI |
There was a problem hiding this comment.
Not for this PR, but: Given we go to the trouble of reading leafCSpWt from the param file, why does it get combined with a constant leafC <- 0.48 instead of reading cFracLeaf from the very next param line?
| c("plantWoodInit\t30000\t0\t6600\t14000\t200", | ||
| "leafCSpWt\t32\t0\t13\t500\t0"), |
There was a problem hiding this comment.
Sipnet v2 drops the extra numeric columns and I would hope it's safe to do that here too:
| c("plantWoodInit\t30000\t0\t6600\t14000\t200", | |
| "leafCSpWt\t32\t0\t13\t500\t0"), | |
| c("plantWoodInit\t30000", | |
| "leafCSpWt\t32"), |
There was a problem hiding this comment.
Also: is the plantWoodInit line needed for this test to work?
There was a problem hiding this comment.
no, plantWoodInit is not needed. model2netcdf.SIPNET only reads leafCSpWt from the param file. I will trim sipnet.param down to just that one line
| nc_file <- file.path(real_outdir, "2002.nc") | ||
| expect_true(file.exists(nc_file)) | ||
|
|
||
| nc <- ncdf4::nc_open(nc_file) | ||
| on.exit(ncdf4::nc_close(nc), add = TRUE) | ||
| vars <- names(nc$var) |
There was a problem hiding this comment.
Tests like this seem like a great use case for the netcdf2df() I proposed in #3788 (which isn't merged yet, so no foul for not using it here!) -- one line to slurp the test file into a df instead of needing to remember these incantations every time.
| n2o <- as.numeric(ncdf4::ncvar_get(nc, "N2O_flux")) | ||
| ch4 <- as.numeric(ncdf4::ncvar_get(nc, "CH4_flux")) | ||
| gpp <- as.numeric(ncdf4::ncvar_get(nc, "GPP")) |
There was a problem hiding this comment.
All of these should be stored as numeric already, and if they aren't then we want to know (as the expect_equal below will do if we skip coercion)
| n2o <- as.numeric(ncdf4::ncvar_get(nc, "N2O_flux")) | |
| ch4 <- as.numeric(ncdf4::ncvar_get(nc, "CH4_flux")) | |
| gpp <- as.numeric(ncdf4::ncvar_get(nc, "GPP")) | |
| n2o <- ncdf4::ncvar_get(nc, "N2O_flux") | |
| ch4 <- ncdf4::ncvar_get(nc, "CH4_flux") | |
| gpp <- ncdf4::ncvar_get(nc, "GPP") |
There was a problem hiding this comment.
right! that these should already be numeric , as.numeric() was masking the type check. But ncvar_get() returns a 1-D array (carries dim attribute), so bare comparison with rep() fails on attributes. Switched to as.vector(), which strips the array dimension without coercing type
c00644b
3 participants
Description
Motivation and Context
Review Time Estimate
Types of changes
Checklist: