Jump to content

 
Photo

Tearing my hairs over NetCDF

- - - - -

  • Please log in to reply
12 replies to this topic

#1
frax

frax

    Hall of Fame

  • Associate Admin
  • PipPipPipPipPipPipPip
  • 2,304 posts
  • Gender:Male
  • Location:Stockholm, Sweden
  • Interests:music, hiking, friends, nature, photography, traveling. and maps!
  • Sweden

gnnnn!!!

Has anyone succesfully used IPCC climate NetCDF data - model outputs in ArcGIS? I have been tearing my head over trying to get something like this to open in ArcGIS - and then looking for alternative ways to get this into a workable state (tiff, ascii grid whatever). I have examined the data in Panopoly, and the data looks ok, and I found an extremely complicated workaround through that software.

If anyone is interested to play with it - http://nomads.gfdl.n...O0.2001-2300.nc - one of the pieces that I tried to view.
Hugo Ahlenius
Nordpil - custom maps and GIS
http://nordpil.com/
Twitter

#2
Hans van der Maarel

Hans van der Maarel

    CartoTalk Editor-in-Chief

  • Admin
  • PipPipPipPipPipPipPip
  • 3,869 posts
  • Gender:Male
  • Location:The Netherlands
  • Interests:Cartography, GIS, history, popular science, music.
  • Netherlands

Well...

It's supported (reading) by FME Professional and up. It appears there needs to be a TAB file for georeferencing as well and it looks as though it supports more bands than an Arc Ascii grid, but that's nothing that can't be fixed...
Hans van der Maarel - Cartotalk Editor
Red Geographics
Email: hans@redgeographics.com / Twitter: @redgeographics

#3
frax

frax

    Hall of Fame

  • Associate Admin
  • PipPipPipPipPipPipPip
  • 2,304 posts
  • Gender:Male
  • Location:Stockholm, Sweden
  • Interests:music, hiking, friends, nature, photography, traveling. and maps!
  • Sweden

did you take a peek at the file I referred to? (if you have time/energy and are curious?)

I don't worry so much about the georeferencing, I can do that later, when I get it in a format I can handle...

NetCDF is often used for temporal data, and the data I am looking at is that - so each time interval can be seen as a band.
Hugo Ahlenius
Nordpil - custom maps and GIS
http://nordpil.com/
Twitter

#4
mckirk

mckirk

    Newbie

  • Validated Member
  • Pip
  • 8 posts
  • Gender:Male
  • Location:Cascadia
  • United States

It seems that the best way to get NetCDF to grid is to create a feature layer first, (http://forums.esri.c...p...40&t=213677) and then convert it to raster. Be sure that you set the Row Dimensions to lat and long, or you will get just one point. I've attached the shapefile that I generated.

Good luck.

Attached Files

  • Attached File  sic.zip   738.86KB   29 downloads


#5
mdsumner

mdsumner

    Key Contributor

  • Validated Member
  • PipPipPip
  • 96 posts
  • Australia

I would recommend using gdal_translate in FWTools to extract the layers or subdatasets you need into some sensible format. You can fix up any georegistration issues with that if needed.

#6
mdsumner

mdsumner

    Key Contributor

  • Validated Member
  • PipPipPip
  • 96 posts
  • Australia

Actually that won't work, the raster is not on a regular grid in Y, and there's no control points for gdalwarp to use.

The route involving interpreting it as a vector layer is probably your best bet, but certainly gdal will translate it into a usable GeoTIFF or whatever you like (it does have subdatasets so you need to use them in the file name argument) and handle the non-regular Y in some way afterwards. (Some netcdf files I've seen have lon/lat stored but really it's a regular Mercator grid, that doesn't seem to be the case here: see the attached plot of the latitudes). You might be able to use the gdalwarp route with control points set up manually, but I'm not familiar with that.

I'd follow up the NOAA site to see how they use it, but if you aren't using Matlab or R or something that can georeference a non-regular "grid" then the vector workaround is probably all you have.

Posted Image

#7
frax

frax

    Hall of Fame

  • Associate Admin
  • PipPipPipPipPipPipPip
  • 2,304 posts
  • Gender:Male
  • Location:Stockholm, Sweden
  • Interests:music, hiking, friends, nature, photography, traveling. and maps!
  • Sweden

MVP points to mckirk - that did it! Thanks, I wasn't sure on how to use that Arc tool to display it, and "Row dimensions" seemed like a black hole to me, I had no idea what to put in there...

But... I seem to have follow up question - the values seem to be the same regardless of what I put in for time...
Hugo Ahlenius
Nordpil - custom maps and GIS
http://nordpil.com/
Twitter

#8
mckirk

mckirk

    Newbie

  • Validated Member
  • Pip
  • 8 posts
  • Gender:Male
  • Location:Cascadia
  • United States

I confess that I'm entirely ignorant of what the data represent. Is "sic" a measure of presence of sea ice? Anyway, when I fiddled with the Time Dimension settings in Layer Properties, I did see some variation in the "sic" values in the polar regions. I've attached a crude animation of the change with frames at 2001, 2015 and 2030.
Attached File  sic_anim.gif   62.35KB   54 downloads

#9
frax

frax

    Hall of Fame

  • Associate Admin
  • PipPipPipPipPipPipPip
  • 2,304 posts
  • Gender:Male
  • Location:Stockholm, Sweden
  • Interests:music, hiking, friends, nature, photography, traveling. and maps!
  • Sweden

sic = sea ice concentration - I will look at it again now!
Hugo Ahlenius
Nordpil - custom maps and GIS
http://nordpil.com/
Twitter

#10
frax

frax

    Hall of Fame

  • Associate Admin
  • PipPipPipPipPipPipPip
  • 2,304 posts
  • Gender:Male
  • Location:Stockholm, Sweden
  • Interests:music, hiking, friends, nature, photography, traveling. and maps!
  • Sweden

I can't seem to get the time dimension to display properly...

I can see the date in the layer properties, and I have it set to the maximum: "2300-04-21 12:00:00" but it still displays the same point values as for "2001-07-02 12:00:00" (default/minimum)

This is what it looks like (see attachment)

Attached Files


Hugo Ahlenius
Nordpil - custom maps and GIS
http://nordpil.com/
Twitter

#11
mckirk

mckirk

    Newbie

  • Validated Member
  • Pip
  • 8 posts
  • Gender:Male
  • Location:Cascadia
  • United States

Hmm... looks right to me. I tried the same thing, clicked OK and got different concentration values around the poles (with an average decrease of 12.74). Maybe there's a problem with your feature layer. I used the following settings when generating the layer:

Variables: sic
X Variable: lon
Y Variable: lat
Row Dimensions: lon, lat
Dimension Values: time

#12
frax

frax

    Hall of Fame

  • Associate Admin
  • PipPipPipPipPipPipPip
  • 2,304 posts
  • Gender:Male
  • Location:Stockholm, Sweden
  • Interests:music, hiking, friends, nature, photography, traveling. and maps!
  • Sweden

mckirk - I used those same settings. Do you get the dates presented in iso format( yyyy-mm-dd hh-MM) or in any other format (dd/mm/YYYY for instance). I have a vague feeling that there could be a locale bug in ArcGIS...
Hugo Ahlenius
Nordpil - custom maps and GIS
http://nordpil.com/
Twitter

#13
mckirk

mckirk

    Newbie

  • Validated Member
  • Pip
  • 8 posts
  • Gender:Male
  • Location:Cascadia
  • United States

I imagine that you've already confirmed this yourself, but you're right! After I modified the regional options on my machine, changes in the Time Dimension produced no change in the concentration values. Apparently, ArcMap isn't clever enough to match regional date formats to the format in the NetCDF. The default regional date settings (in WinXP) for the U.S. will work:

M/d/yyyy
h:mm:ss tt




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users

-->