**********************************************************************************************************
The input data to sim_obs_conv.f90 and sim_obs_rad.f90 are extracted from the original NR grib data.  This
note describes the data structure and format.

Note: Please let me know if you have any questions or confusing.  I have
tried to include all things and to give a clear description.  But, it is not surprising
that I could overlook a thing.  Thanks for your questions and comments in advance.

Writen by Runhua Yang at GMAO/NASA/GSFC 
Email: runhua.yang-1@nasa.gov, Tel: 301-614-5251
August 20, 2008

*****************************************************************************************************
1. General description of the input binary data files
-----------------------------------------------------

Input data are extracted from the EC Nature Run with T511 resolution.  The grid is Gaussian grid
with 91 vertical hybrid levels. Our data are provided by Dr. Arlingdo da Silva at GMAO. 

The extracted fields include five 3-d fields and two files of surface fields. The full list of
the fields and the associated variable strings are listed in section 1.2.  The data interval is 3 hours. 
The sim_obs_conv.f90 and sim_obs_rad.f90 require the input data in the way as described below.  Please
read our document about the software about these two programs.
If you use our scripts to extract the data or do simulating OSSE data, please use
the same naming convention as given in the job scripts.  

2 Data structure and format
-----------------------------

The input data files for a record, say, Jan. 1st, 12z, 2006, are:

odat_2006010100z.bin  -ozone in unit: Ozone mass mixing ratio [kg kg**-1]
qdat_2006010100z.bin  -specific humidity in unit [kg kg**-1]
tdat_2006010100z.bin  -temperature  in unit K
udat_2006010100z.bin  -horizontal winds  meter/s
vdat_2006010100z.bin  -meridional winds  meter/s
pres_2006010100z.bin  -include surface pressure in Pascal and meter
surf_2006010100z.bin  -see below about the variables included.

Note that NR reports the Logarithm of surface pressure and geopotential height, but sim_obs_conv.f90 
and sim_obs_sat.f90 need surface pressure in Pascal unit and height in meter.  

I put whole set of extracted Nature Run data in this ftp site.  They are in little-endian form.  You
may give a try to run sim_obs_conv.f90 with them. 

The format of all these input files are given by several codes.  I will give 
a separate document to describe these codes and associated job scripts soon, hope by the end of
today.  You do not need them at this moment since you can try to use the given
sample data in this site to run sim_obs_conv.f90.

2.1 Format for 3-d fields 
--------------------------
Formats for all five 3-d fields are similar. I give the written 
format with them these binary data are created. Similarly, the format
of other files are the written format.  The machine creates little-endian.

Note: the open format of all binary files are the same as:
open (unit,file='xxx',access='sequential',form='unformatted')

format for 3-d fields:
write (unit) vchar,rlev,(r4data(ii),ii=1,348528)

Here, r4data are real*4 and the length is 348528. Data start from
90N.  The starting longitude is zero.

"vchar" is a variable string, rlev denotes which vertical level out of 91 is written out.
A full list of the 3-d fields strings is:
       uchar = 'uwnd'
       vchar = 'vwnd'
       tchar = 'temp'
       qchar = 'sphu'
       o3char= 'ozon'

2.2 Format for surface fields 
--------------------------------

Data are created with the written format:
       write(unit) varchar(nn),nn,(r4data(ii),ii=1,348528)

Similarly, varchar is a variable string, nn is an integer denote
which field among the 12 surface fields, listed below.  r4data denotes a real*4 fields.


The variable strings are listed as:
      data varchar /'ismk','rain','conp','u10m','v10m','tp2m',
     1 'dt2m','almk','lcld','mcld','hcld','sktp'/

The description of these variables are:
c var characters for different fields:
c  ismk: surface sea-ice cover (0,1)
c  rain: surface stratiform precip. (m)
c  conp: surface convective precip. (m)
c  u10m: uwind at 10 meter
c  v10m: vwind at 10 meter
c  tp2m: surface temperature at 2 meter
c  dt2m: dew point temperature at 2 meter
c  almk:  sea (Aqua) land mask (0,1)
c  lcld: low cloud cover fraction (0,1)
c  mcld: middle  cloud cover fraction (0,1)
c  hcld: high cloud cover fraction (0,1)
c  sktp:  surface skin temperature  (k)
c......................................

2.3 Format for surface pressure and surface height
--------------------------------------------------
for surface height:
 write(50) varchar,1,(r4data(ii),ii=1,348528)
The varchar is:
  varchar = 'zsfc'
and for surface pressure:
write(50) varchar,2,(r4data(ii),ii=1,isec4(1))
varchar is:
 varchar = 'pres'

3. Fortran code and shell scripts 
---------------------------------
  I will put it in this ftp site by the end of today.