Re: satellite data

[Jonathan Callahan <Jonathan.S.Callahan@xxxxxxxx>]

Jean-Francois,

By default, LAS tells Ferret to create graphics using the Ferret 'FILL'
command which smoothes the data by interpolating.  When there are a lot of
missing values this sometimes results in unattractive results.  You can use
the 'Output Options' page to request that the graphic use 'Contour fill style
= raster'.  This results in Ferret using a 'SHADE' command to plot the
colored map.   This should give an accurate representation of the data and
may look a whole lot better for swath data.

You can specify which graphic properties to use by default on a per variable
basis in the dataset configuration files:

http://ferret.pmel.noaa.gov/Ferret/LAS/FAQ/per_variable_configuration.html

I would agree with your assessment, however, that Ferret and LAS need some
improvements in the realm of satellite data support.  Unfortunately, we've
never been funded to explore the possibilities.


-- Jon


Jean-Francois PIOLLE wrote:

> Hi Jonathan,
>
> thank you for your answer about time series : I am now trying to set up
> a Grads-Dods server to serve my maps (input to LAS); I will keep you
> informed of the added value compared to the DODS aggregated server.
>
> My goal is actually to serve satellite data. Here is my first experience
> :
>
> - all our data are in netCDF or HDF, and I think this is a common rule
> (for instance at JPL/PO.DAAC)
> - I plan to serve level 3 products but of 2 kinds :
>    * analyzed products : interpolated grids, with various time
> resolution (one every 3-hour, 12-hour, day, week, month, etc...); This
> is why I sent this posting about time series : everything is working
> fine except this performance problem when accessing for the first time
> to a dataset served by an aggregated DODS server (even for an xy plot).
>    * swath products : this product is close to level 2 product except
> that the satellite swath is mapped on a regular grid (could be the world
> or a specific region). We have of course the same problem than above
> (long time series) but also graphic plot with ferret appears to be very
> problematic :  since a lot of values are missing (outside the swath), I
> think the graphical  interpolation does not work properly (for ex. with
> the xy plot) : the output gif is sometimes very strange and Ferret just
> plot a few patterns of the swath (but never the complete swath). I guess
> a specific custom plot is  needed in that case (may be it already
> exists?)
>
> I would like to serve level 2 data (that we archive here at CERSAT) but
> I don't think it is possible without the use of an "intelligent"
> database keeping an index of each orbit file with georeferencing
> metadata in order to retrieve quickly the list of the orbit files
> available over a given area in a given time period.
>
> As a conclusion, there is really a need to bring some improvements to
> LAS to serve satellite data (apart of interpolated fields which do not
> differ from model outputs); we are just starting our ivestigation with
> LAS problematic so we have very few experience to share (except the
> problems we face) but as soon as we solve some issues, I will keep you
> informed. Any experience or advice are also welcome.
>
> Jean-Francois
>
> Jonathan Callahan wrote:
> >
> > Hi all,
> >
> > The response to my last posting about time series was so informative
> > that I'd like to seed the group with another, more general question:
> >
> > Can anyone offer pointers about working with satellite data?
> >
> > We have some new contacts considering adopting LAS/Ferret/OPeNDAP for
> > working with level 1, 2 and 3 satellite data and I have precious little
> > advice to give them.  Any suggestions about formats, translating
> > servers, alternative back ends, etc. would be most helpful.
> >
> > -- Jon
>
> --
> -------------------------------------------------------------
> Jean-Francois PIOLLE
> CERSAT  French ERS Processing and Archiving Facility
> IFREMER
> BP70
> 29280 Plouzane
> FRANCE
>
> Tel.:  (+33) 2 98 22 46 91   email: jfpiolle@ifremer.fr
> Fax:   (+33) 2 98 22 45 33   WWW:   http://www.ifremer.fr/cersat
>
> -------------------------------------------------------------