Commits

Author Commit Message Labels Comments Date
Tom Roche
README TODOs tweaked Added note to use bash booleans à la https://bitbucket.org/tlroche/aqmeii-na_n2o_integration/src/HEAD/N2O_integration_driver.sh?at=master after redoing integrations to step up to new AQ emissions Dec 2013. Also standardized * proper case, punctuation * quoting of code references ("proper nouns") Not tested: doc not code.
Tom Roche
README.md tweak: add reminder to retest with latest regrid_utils
Tom Roche
README.md: add TODO for regrid_utils diff's found with regrid_utils/repo_diff.sh, tweak URI to sources
Tom Roche
tweak
Tom Roche
added "standard" TOC and section=TODOs, switched github references to bitbucket (live ones, anyway)
Tom Roche
forgot to re-`git add` last few changes to README
Tom Roche
get onto latest regrid_utils, use for filepath templating Main difference is using regrid_utils::get_filepath_from_template.ncl consistently for that task, which means * single template string='@@@@@@@@' * always pass/replace year * (in NCL) paying attention to types of {year, month, day} tokens: regrid_utils::get_filepath_from_template.ncl wants int arguments (which even affected vis_regrid_vis.r, except for latter). Also moved TODOs from these commit…
Tom Roche
allow uber_driver to work with GFED_DOWN and ACCESS != file (and harmonize truth value). TODO: test from uber_driver:file
Tom Roche
update README after successful completion of 'env GIT_SSL_NO_VERIFY=true ./GFED_driver.sh' on terrae
Tom Roche
fix grid alignment, fully IOAPIize hourly output, enable view in VERDI After fully IOAPI-ize output from make_hourlies.ncl, and viewing in VERDI (which also verifies goodness of IOAPI-ization), I was able to * verify that make_hourlies.ncl's output was flipped (just like the previous projects') * fix it Also: * add lots debugging and progress code to make_hourlies.ncl * add lots progress code to check_conservation.ncl (more below) * removed no-longer-used output co…
Tom Roche
get fully on regrid_utils ... preparatory to writing fully-IOAPI-zed output Apparently never fully got GFED_driver.sh and callees using regrid_utils fully, so whacked * duplicate functions * old-style NCL library loads * use of get_daily_emis_fp.ncl (just now removed from regrid_utils) Tested on terrae/EMVL from 'uber_driver.sh' in fresh terminal: didn't wait for check_conservation.ncl to finish, since a broken forward hosed gmail making firefox swap and threatening to c…
Tom Roche
adopts current *_driver framework preparatory to writing fully-IOAPI-zed output uber_driver.sh and GFED_driver.sh resemble peers from other subprojects, including * uber_driver failfasts * use of AQMEII_ag_soil output for template (not the CMAQ kludge previously used only for extents) Also: * tweaks to vis_regrid_vis.r, check_coord_vars.ncl (to accommodate above) * added strings to GFED_driver.sh preparatory to fully-IOAPI-zing output Tested on terrae/EMVL from * 'env GIT_S…
Tom Roche
more TOC fiddling
Tom Roche
Tom Roche
works around down zea.ess.uci.edu, almost fully ported to AQMEII_ag_soil-style framework + uber_driver.sh and GFED_driver.sh support use of local archives for daily and 3-hourly fractions + pulls all non-project-specific utilities from regrid_utils + does AQMEII_driver-style failfast: GFED_driver, uber_driver - does not produce IOAPI-compliant output (much less try to launch VERDI) Also: * tweaks to vis_regrid_vis.r, make_hourlies.ncl * check_conservation.ncl uses summarize.ncl (and …
Tom Roche
(not fully tested due to long runtime on terrae) fully ported to new-style framework * moved bash_utilities.sh, string.ncl, time.ncl, visualization.r to regrid_utils * GFED_driver.sh (esp ::get_helpers and callees) uses regrid_utils * GFED_driver.sh env var names, and function names and forms, more consistent with GEIA_driver, CLMCN_driver, EDGAR_driver * README more consistent with those from GEIA, CLMCN, EDGAR * regrid_global_to_AQMEII.r uses new-style visualization calls * *.nc…
Tom Roche
kg -> mol in docs (code is already mol)
Tom Roche
Added check_conservation.ncl to check conservation of mass GFED input -> hourly output. Note check_conservation.ncl uses string.ncl (refactored from other projects). check_conservation.ncl output on terrae (excerpted) > GFED monthly N2O (mol) over globe from ./GFED-3.1_2008_N2O_monthly_emissions.nc: > |cells|=3.1104e+06 > |obs| =3.1104e+06 > min =0.000e+00 > q1 =0.000e+00 > med =0.000e+00 > mean =5.443e+03 > q3 =0.000e+00 > max =4.440e+07 > sum =1.693e+10 > m…
Tom Roche
make_hourlies.ncl: do not recreate output if isfilepresent(fp) (note odd NCL "end if" semantics!)
Tom Roche
Ensured (hopefully) output in units=molN2O/s as needed for CMAQ input. ... and better documented the conversion so that I don't forget I had already done most of it, like I did yesterday :-( Also: * refactored separate get_daily_emis_fp.ncl for use in other projects. * did some setup for conservation checking (notably, check_conservation.ncl) since I discovered the problem above when coding for that. * created get_output_areas.ncl to solve a problem I didn't actua…
Tom Roche
testing from uber_driver pulling from repo found bug, hopefully now fixed; updated README
Tom Roche
Add make_hourlies.ncl to create final-output hourly emissions (in daily files). make_hourlies.ncl enacts the GFED protocol for generating 3-hourly emissions (from ftp://gfed3:dailyandhourly@zea.ess.uci.edu/GFEDv3.1/Readme.pdf ) then linearly downscales (/3 :-) to hourly, writing (hopefully) IOAPI-compliant output. Also: * added first_day_of_month(...), last_day_of_month(...) to time.ncl for use in daily-emission loop. * minor tweak to fix_3hourlies.ncl Tested on terrae/…
Tom Roche
add check_coord_vars.ncl check_coord_vars.ncl checks the presence and orientation (i.e., values increasing or decreasing) of coordinate variables in major products: * monthly emissions and regrids * daily fractions and regrids * 3-hourly fractions and regrids * extents template Tested on terrae/EMVL from 'GFED_driver.sh', and only in isolation, since this is not currently only an optional part of the main driver loop. …
Tom Roche
integrate fix_3hourlies.ncl into drivers fix_3hourlies.ncl recombines the 3hourly regrids produced by R::raster (and fixes values(LAY), and twiddle some attributes, producing > Variable: temp_var > Type: float > Total Size: 52700544 bytes > 13175136 values > Number of Dimensions: 4 > Dimensions and sizes: [TSTEP | 12] x [LAY | 8] x [ROW | 299] x [COL | 459] > Coordinates: > TSTEP: [1..12] > LAY: [0..2…
Tom Roche
re-regrid 3hourly fractions so as to have dim=LAY in output Could not figure out how to make R::raster regrid a 4D variable, so split it on TSTEP and regridded the 3D pieces (LAY, ROW, COL). Noticed one problem: projectRaster seems to have changed the values of LAY from [0,3,...,21] to [1,...,8]. Tested on terrae/EMVL from both 'GFED_driver.sh' and 'uber_driver.sh' (latter in fresh terminal). TODO: * use NCL to recombine the 3hourly regrids (and fix LAY,…
Tom Roche
fix deadly brainfart in GFED_driver.sh
Tom Roche
make visualization routines/sequences more consistent Added visualization.r to this repository to separate visualization.r::visualize.{layer,layers} -> visualization.r::{plot,nonplot,visualize(==nonplot+plot)}.{layer,layers}. Now can use nonplot-ers with fractions {conversion, regrid} for more consistent presentation, which hopefully will allow me to notice problems like missing LAY more quickly. Tested on terrae/EMVL from 'GFED_driver.sh' TODO: * …
Tom Roche
global namespace cleanup for sanity Better differentiate {daily, 3hourly} fractions and emissions, now that I'm about to create the latter: phase 1 GFED_N2O_ -> GFED_GLOBAL_ GFED_N2O_...REGRID -> GFED_REGRID_ phase 2 (depends on phase 1) GFED_N2O_DAILY_REGRID_ -> GFED_REGRID_DAILY_FRAC_ to differentiate from the GFED_REGRID_DAILY_EMIS_ I'll be creating GFED_N2O_3HOURLY_REGRID_ -> GFED_REGRID_3HOURLY_FRAC_ (parallel motivation) phas…
Tom Roche
update README with links to uploaded plots
Tom Roche
fixed problem regridding fractions Discovered - I had not completed vis_regrid_vis.r: while I both visualized (plots and stats) and regridded monthlies, I only regridded the fractions (dailies and 3hourlies) - on visualizing the regridded fraction files, I discovered their structure was incorrect, and they had max=Infinity While doing that, discovered that netCDF.stats.to.stdout.r could not handle a true 4-dimensional array (i.e…
  1. Prev
  2. Next