U.S. JGOFS Synthesis & Modeling Project

USJGOFS SMP Research Working Groups Updates Contacts Reprints Data

DATA:  Guidelines for Preparing Gridded Data Sets for the LAS

data bar

Inventory Data SMP LAS Guidelines Resources Data Policy Survey Links

For gridded dataset we strongly recommend that investigators supply their data in netCDF format and adhere to the COARDS and GDT conventions.

  • NetCDF is a widely used, self describing, machine independent format that allows for array oriented data subsetting.
  • It is both free and well supported by Unidata.
  • Software libraries for C, Fortran, C++ and perl are available.
  • More information on netCDF, the netCDF library and other netCDF tools is available at: http://my.unidata.ucar.edu/content/software/netcdf/index.html
The following 3 questions will help identify any special handling required for your data. If you answer YES to any of these questions, please contact the U.S. JGOFS Data Management Office via request form.
1. Is the grid lon-lat-depth-time curvilinear? (i.e. not rectilinear)
Posting curvilinear grids on the LAS system requires special handling
2. Are there special visualization requirements? (e.g., palette, masking, levels)
An example of this is where your data requires displaying on some non-linear scale.
3. Does the dataset consist of multiple files that need to be strung together?
1. Specify which variables from this dataset should be (can be) displayed with LAS
2. Every variable must have the following three attributes:units, missing_value and long_name.
3. Although LAS doesn't require it, the global attributes should include: description, authors and date.
4. If it's a global data set, LAS also expects the longitude axis to have: LON:modulo = " " ;
If you are familiar with netCDF or Ferret,the following additional checks might also prove useful:
ncdump -h ~file~
1. Do the dimensions appear again as variables?
2. Are the axis units defined? Are they ones that Ferret understands?
3. Do the data variables have reasonable titles and units?
4. Is the missing value flag specified?

start FERRET and look at the data:

1. USE ~file~
  • can ferret read the data?
  • are the variables and their units reasonable?
  • are the grid dimensions sensible?
  • are the modulo axes understood?
4. SHADE ~var~
  • does the plot look reasonable?
  • does it need a land mask?
  • is it right side up?
  • are the missing values understood?
METADATA EXAMLE: An example of what LAS considers adequate metadata is shown below. This is the 'ncdump -h filename' listing for the Global SeaWiFS Chlorophyll data currently listed on the SMP LAS, in the Ocean Color category.

ncdump -h SeaWiFS_data.nc

    netcdf SeaWiFS_data {
          LONGITUDE = 360 ;
          LATITUDE = 180 ;
          TIME = 138 ;
          float LONGITUDE(LONGITUDE) ;
             LONGITUDE:units = "degrees_east" ;
             LONGITUDE:long_name = "Longitude" ;
             LONGITUDE:point_spacing = "even" ;
          float LATITUDE(LATITUDE) ;
             LATITUDE:units = "degrees_north" ;
             LATITUDE:long_name = "Latitude" ;
             LATITUDE:point_spacing = "even" ;
          double TIME(TIME) ;
             TIME:units = "days since 1950-01-01 00:00:00" ;
             TIME:long_name = "Time" ;
             TIME:time_origin = "1-JAN-1950" ;
             TIME:point_spacing = "uneven" ;
             CHLOROPHYLL:units = "mg/m^3" ;
             CHLOROPHYLL:long_name = "Chlorophyll" ;
             CHLOROPHYLL:missing_value = "1e+35" ;
       // global attributes:
             :Title = "Global SeaWiFS chlorophyll 8 day composites" ;
             :authors = "J. Yoder & M. Kennelly" ;
             :date = "26 July, 2001" ;


These guidelines were kindly drafted by Jon Callahan of  NOAA/PMEL, and Cyndy Chandler of  WHOI
Draft: 29 Jan 2001
Updated: 17 Sep 2001