20040408: national mosaic (radar)

Greg,

The GINI National Mosaic in the IDD stream with feedtype FNEXRAD
are the following:

1km: N0R product composite      rad/NEXRCOMP/1km/n0r_20040408_2111
2km: N1P product composite      rad/NEXRCOMP/2km/n1p_20040408_2112
4km: NTP product composite      rad/NEXRCOMP/4km/ntp_20040408_2114
                                                     YYYYMMDD HHNN

The format of the GINI header is described in the NESDIS ICD:
http://www.nws.noaa.gov/noaaport/document/icd_ch4.pdf

The products are created in lamber conformal, eg table 4.4A in the
above PDF. The Actual GINI header is preceeded by a 21 byte WMO format 
header. That plus the 512 bytes of GINI header give 533 bytes, followed
by the PNG format product.

We have a program which will unpack the PNG into the raster if you
desire:
pngg2gini in the ldm-mcidas.tar.Z package you can obtain from
the pub/binary directory of ftp.unidata.ucar.edu.

Or, our GEMPAK distribution will read the PNG directly, or as
uncompressed into the standard raster (byte 43 in the GINI header
signals compression....I am using a value of 128 to signal that PNG is
used for the raster). You can also just pull out the PNG and
display it directly from the 522 byte offset.

You will find the following information from the GINI header:
Nx = 4736
Ny = 3000
Proj = LCC with stdlon = -100 and tangent lat = 40.

For the products, I have the band values defined as:

!Prod   band ncal units,Calval_1;....;calval_n  
N0Z       25    1 dBZ,0,255,0,75            
NET       26    1 KFt,0,255,0,85            
N0R       27    1 dBZ,0,105,-30,75           
NCR       28    1 dBZ,0,105,-30,75           
NVL       29    1 Kg/m**2,0,255,0,85         
N1P       30    3 inches,10,110,0,2;110,210,2,6;210,255,6,12   
NTP       31    3 inches,10,110,0,4;110,220,4,12;210,255,12,24     

The calibration information is signaled with byte 47 of the header=128
and I described the format:
http://www.unidata.ucar.edu/projects/coohl/mhonarc/MailArchives/gempak/msg04994.html

For the n0r, the calibration is simply byte values 0 to 105 represent
dbz values of -30 to 75. 

In the case of N1P, there are 3 segments of the calibration:
values 10 to 110 scale to 0 to 2 inches
values 110 to 210 scale to 2 to 6 inches
values 210 to 255 scale to 6 to 12 inches

The non linear calibration (eg 3 linear segments) is based on the fact 
that the NIDS products have discrete values, and the bin values are
not linearly spaced.


The GRIB products currently are 6km national and 1km regional in the
FNEXRAD feed using product IDs:
 radar_mosaic_national !grib/unidata/1/#255/200404082126/F000/N0R/sfc!     
000000
 radar_mosaic_regional_ilx !grib/unidata/2/#255/200404082126/F000/N0R/sfc! 
000000

The #255 is the GDS defined grid number. These use a model center id 60
(eg NCAR/UCAR) with a subcenter id of 1 (my self defined unidata 
subcenter). The regional product is centered on the station id in the
product name (and is relocatable by one of our universities).

The reason for the 6km grib for national was a limitation on the user
end to handle the number of grid points in a 1km product.

I have a web page set up that describes the products:
my.unidata.ucar.edu/content/software/gempak/examples/gdradr
http://motherlode.ucar.edu/unidata/images/nids/gdradr.html


Steve Chiswell
Unidata User Support


On Mon, 2004-04-05 at 20:44, Greg Thompson wrote:
> Steve,
> 
> a number of months ago you mentioned the existence of a national
> radar mosaic product you created.  Can you tell me the header
> to pick it up over LDM?  I recall that the product is 533 bytes of
> header data followed by a PNG image.  Can you tell me more about the
> projection information and anything I can glean from the header?
> Also is it little-endian or big?
> 
> I'm working to replace Peter Neilley's software that creates NIDS
> single-site images as well as his display of the WSI-Nowrad product
> since that has restrictions placed upon it.  RAP can no longer support
> his software and I've replaced all his single-site stuff with
> improved images.  Now, I'm looking to re-map sectors of your hi-res
> mosaic into regions around the country (probably the same sectors
> as seen on my satellite and surface data pages).
> 
> Also, you mentioned you were planning to GRIB a radar mosaic product.
> That item had 6-km spacing whereas the PNG file was 1-km from what
> I recall.  I'd truly love to get the 1-km data as a GRIB file so I
> could manipulate any way I desire but I'll take what's available now
> if you can point me in the right places.
> 
> Thanks!