predictiveecology / biomass_speciesdata Goto Github PK

per LandWebProprietaryData, the "new" way is to use sim$rasterToMatch as we need more than just a projection for the rasters

As pointed out in #32 , we should ideally be using Plots to plot the final speciesLayers. Switching to Plots can be done at the module level or in LandR::plotVTM

input objects like e.g. studyArea and studyAreaLarge, which are shared across LandR modules need to always have the same defaults in .inputObjects

We need to change httr::config throughout this and other modules to be able to download KNN layers.
Solution:

• create a new module parameter that controls httr::config(ssl_verify) and wrap problematic prepInputs calls into httr::with_config()

This module should be generic enough to be converted to an R package.

I really want to get the Initial Types plot produced as png, so improving the plotting in this module is needed.

transfer to eliotmcintire (eventually, will be at PredictiveEcology but we aren't ready for that to be public there yet)

the final [filebacked] raster stack that is output from this modules needs be stored in the user's directory rather than a temporary location. all these temp files are clogging up storage space, and may not persist between reboots.

this likely goes deeper than just this module (i.e., likely needs to be addressed in prepInputs et al.)

please subscribe the slack channel to updates when the change is complete

SpaDES packages other than SpaDES.addins are being installed when install.packages("SpaDES", dependencies = TRUE) is run in the GH workflow:

trying URL 'https://packagemanager.rstudio.com/cran/__linux__/focal/latest/src/contrib/SpaDES.core_1.1.0.tar.gz'
Error in download.file(url, destfile, method, mode = "wb", ...) : 
  cannot open URL 'https://packagemanager.rstudio.com/cran/__linux__/focal/latest/src/contrib/SpaDES.core_1.1.0.tar.gz'
In addition: Warning message:
In download.file(url, destfile, method, mode = "wb", ...) :
  URL 'https://rspm-sync.rstudio.com/bin/4.2-focal/[428](https://github.com/PredictiveEcology/Biomass_speciesData/actions/runs/3252427112/jobs/5338629347#step:7:429)ad5ba346e0b5ca56fbf85c942d0492880c1306ef36070df1ce490b2b15293.tar.gz': status was 'SSL connect error'
Warning in download.packages(pkgs, destdir = tmpd, available = available,  :
  download of package ‘SpaDES.core’ failed
trying URL 'https://packagemanager.rstudio.com/cran/__linux__/focal/latest/src/contrib/SpaDES.tools_1.0.0.tar.gz'
Error in download.file(url, destfile, method, mode = "wb", ...) : 
  cannot open URL 'https://packagemanager.rstudio.com/cran/__linux__/focal/latest/src/contrib/SpaDES.tools_1.0.0.tar.gz'
In addition: Warning message:
In download.file(url, destfile, method, mode = "wb", ...) :
  URL 'https://rspm-sync.rstudio.com/bin/4.2-focal/4a3ea2cd0123b232eca27250dd0bc622e898645bd67578ef535a7397aa39499f.tar.gz': status was 'SSL connect error'
Warning in download.packages(pkgs, destdir = tmpd, available = available,  :
  download of package ‘SpaDES.tools’ failed
trying URL 'https://packagemanager.rstudio.com/cran/__linux__/focal/latest/src/contrib/covr_3.6.1.tar.gz'
Content type 'binary/octet-stream' length 322914 bytes (315 KB)
==================================================

similar to the AOU/ROF species hists produced by Benoit

• to avoid downloading all species layers.
• LandR::speciesInStudyArea must be able to deal with both kNN 2001 and 2011

similar to PredictiveEcology/Biomass_borealDataPrep#22, does the rasterToMatch really need an FRI, or is this a product of copy/pasting from LandWeb?

As per PredictiveEcology/actions#1

Using .plots = c("object", "png", "raw") in headless session (w/o using xvfb) fails with the following cairo graphics error:

Oct20 09:03:00 Biomass_speciesData    total elpsd: 6.2 secs | 0 Biomass_speciesData ini
2022-10-20 09:03:00 WARNING::message call unable to open connection to X11 display '' x11()
2022-10-20 09:03:00 ERROR::e: unable to start device X11cairo
 2022-10-20 09:03:00 ERROR::e: .External2(C_X11, d$display, d$width, d$height, d$pointsize, d$gamma, d$colortype, d$maxcubesize, d$bg, d$canvas, d$fonts, NA, d$xpos, d$ypos, d$title, type, antialias, d$family, optionSymbolFont(d$symbolfamily))
Error in .External2(C_X11, d$display, d$width, d$height, d$pointsize,  : 
  unable to start device X11cairo

I can presumably get around this by starting my R session using xvfb-run -a R or similar, but this imposes addional system dependencies (which user may not have ability to install) and adds an extra layer to the run requirements of this module, which impacts all LandR Biomass sims.

@CeresBarros I know I prefer cairo graphics in some cases, but are they essential here?

prepInputs() call on Pickell dataset generates warnings:

Warning messages:
1: In projectInputs.Raster(x = new("RasterLayer", file = new(".RasterFile",  :
  This raster layer has integer values; it will be reprojected to float. Did you want to pass 'method = "ngb"'?
2: In .getFormat(filename) :
  extension .dat is unknown. Using default format.

an on the CASFRI data generates:

Warning message: In projectInputs.Raster(x = new("RasterLayer", file = new(".RasterFile",  :   This raster layer has integer values; it will be reprojected to float. Did you want to pass 'method = "ngb"'?

The following objects are created in the init event but are listed as inputs in metadata:

• Pickell
• CASFRIRas

They should either be moved to .inputObjects, or should be removed from metadata entirely as they are transient objects used for the creation of sim$speciesLayers.

Browse[1]> uniqueKeepSp
[1] "Abie_sp"  "Pice_gla" "Pice_mar" "Pinu_sp"  "Popu_tre"
Browse[1]> sim$speciesLayers
class       : RasterBrick 
dimensions  : 727, 1070, 777890, 5  (nrow, ncol, ncell, nlayers)
resolution  : 250, 250  (x, y)
extent      : -1005571, -738070.6, 7159155, 7340905  (xmin, xmax, ymin, ymax)
coord. ref. : +proj=lcc +lat_1=49 +lat_2=77 +lat_0=0 +lon_0=-95 +x_0=0 +y_0=0 +ellps=GRS80 +towgs84=0,0,0,0,0,0,0 +units=m +no_defs 
data source : in memory
names       : Pine, Black.Spruce, Deciduous, Fir, White.Spruce 
min values  :    0,            0,         0,   0,            0 
max values  :  100,          100,       100, 100,          100 

see loadCASFRI() and TODO in biomassDataInit()

studyArea should be created first, then studyAreaLarge copied from studyArea if it is missing.

I noticed some TODOs re: which studyArea to use in the prepInputs calls. I am going to be reviewing this for borealDataPrep, so I may as well do that here too.

At the moment I believe studyArea is preferred for everything except stand age. However, some functions may require additional cropping/masking because of this (e.g., biomass may be needed at both the studyArea and studyAreaLarge extent)