[gdal-dev] guidance on NSIDC Sea Ice concentration data

[Michael Sumner]

Hello, I've put together an example of using NSDIC sea ice concentration
data with GDAL VRT.

In this Github readme I list two example data sets, one for the Northern
and one for the Southern hemisphere where the raw
binary NSDIC .bin files have accompanying VRT files:

https://github.com/mdsumner/NSIDC-seaice

I'd really like to see this format as a GDAL driver, but I don't have the
C++ expertise to make it happen. I can offer help and testing if anyone was
interested in writing the driver, but I would also appreciate any guidance
and kick start for me to learn to do it myself.

The format itself is detailed here, and everything required can be gleaned
from the header - though in the VRT pathway I simply skip the header and
apply the right CRS, orientation, grid size and geotransform:

http://nsidc.org/data/docs/daac/nsidc0051_gsfc_seaice.gd.html#format

The repo includes the minimal .vrt required to read this as a valid GDAL
data set and uses plotting in R with standard R
mapping tools to show that the interpretation is right, at least visibly.

I have these questions:

* what is the best template driver for these data, should it derive from
 GDALPamDataset like JDEM or from RawDataset as ERS does? (I'm way out of
my depth here)
* is it appropriate to use this SRS when the datum is unspecified? I'm not
clear on the pitfalls here and how to advise downstream use

I am competent with building GDAL but I get lost in the steps required to
add a new driver. A dummy example, or some more detailed guidance for how I
would apply the "GDAL Driver Implementation Tutorial" to these data would
be enough for me to flesh out the details.

The projection SRS used are EPSG 3411 (north) and 3412 (south) as listed
here:
http://spatialreference.org/ref/epsg/3411/
http://spatialreference.org/ref/epsg/3412/

I have otherwise written this support directly into the R package raster,
the R source for these files is here in this read-only mirror (raster is on
r-forge):

https://github.com/cran/raster/blob/master/R/nsidcICE.R

This works very well for R, but I'd prefer to see this format in GDAL so
that other languages have the same facility. There are other sea ice
concentration data sources, usually in HDF4 format - and they only need the
geotransform and SRS assigned for use in GDAL - so they need work as well
but not as fundamental as this raw binary format.  This data source is very
extensive, with daily data at 25km resolution every day since 1979 with
only ~30 days total missing, and only a handful with problems in the map
coverage.

Thank you.

Cheers, Mike.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/gdal-dev/attachments/20150724/b587ccd9/attachment.html>