"HIRLAM MODEL DATA SPECIFICATION DOCUMENT"
------------------------------------------
Purpose:
--------
Based on data from DMI-HIRLAM the Danish Meteorological Institute (DMI) is
providing estimates of the zenith total delay (ZTD), the wet dalay (ZWD), the
integrated water vapor (IWV), the local pressure, and other quatities
quantities.
The data are meant for validation of gps derived delays and to assist people
having to determine IWV from GPS delays but having no better access to
auxiliary meteorological information about pressure and temperature.
Original data: -------------- The data come from the numerical weather
prediction (NWP) model DMI-HIRLAM, which is the model used operationally at
DMI. The resolution of the model is 0.15 degrees in both horisontal directions
and it has 40 vertical levels. It covers Europe and areas beyond to the west
and north. The model is run with data assimilation every three hours, output is
available with a resolution of 1 hour. For TOUGH we extract data with a
resolution of 1 hour selecting the shortest possible forecast lengths (which
are then 0, 1, and 2 hours).
GPS stations covered.
---------------------
The goal is to provide NWP HIRLAM data for all GPS stations for
which we in TOUGH have access to GPS data. Currently the extractions
are done for the GPS sites for which data were available in the
COST 716 database early March 2003. The 'current' list can be
found on the ftpserver as met/hirlam/gpslist.dat. Always refer to the
coordinates in the data files when using the data, not the list, as the
list will get updated from time to time.
Update of GPS station coordinates and adding of new stations
------------------------------------------------------------
NOTICE: There is currently no automatic update of the GPS station list. If you
find a station is assigned outdated or wrong coordinates, or you want a new
station added to the list of stations, please send an email to tough@dmi.dk
detailing the request. We need the name, the latitude, longitude, and altitude
of the GPS site (antenna). Remember to specify the reference surface for the
altitude you provide.
Data provided for TOUGH
-----------------------
For each location we provide the surface pressure, the zenith total
delay (ZTD), the zenith wet delay (ZWD), the integrated water
vapor (IWV), both at the HIRLAM surface and at the altitude of the GPS
antenna. Furthermore the 2 meter temperature and the 2 meter humidity are
provided for the HIRLAM orography.
Derivation of the data provided.
-------------------------------
The properties of the atmosphere of the region at a given time are determined
from data extracted from the operational HIRLAM runs made at DMI.
The properties of the atmosphere above each GPS site are calculated firstly by
a horizontal interpolation within the HIRLAM grid. From this vertical profile
the estimates of surface pressure, ZTD, ZWD, and IWV at the HIRLAM surface are
derived, as are T2m and RH, the two meter temperature and relative humidity.
Secondly a vertical transformation is applied to the atmospheric profile at
the GPS site to correct for the offset in altitude between the HIRLAM surface
and the real surface (here = the specified altitude of the GPS antenna). One
should notice that there exist no physically 'correct' way of making this
vertical transformation, only educated attempts based on various reasonable
assumptions. After all one is trying to go beyond the resolution of the model.
We adopt a method described by Majewski (Beitr. Phys. Atmosph., 1985, Vol.
58, p. 147), a method which is widely used for mapping meteorological data from
one model grid to another. This method preserves boundary layer structure and
maintains stability properties of the column.
After the vertical shift of the atmospheric profile estimates of
pressure, ZTD, ZWD, and IWV are derived for the GPS antenna level.
Before doing the numerical integrations the integrals are recast
in terms of the variables of the HIRLAM model. This is done using the
hydrostatic assumption (which is also a key constituent of the HIRLAM
model itself, and a two component ideal gas describtion of the atmosphere
[with the dry atmosphere constituting one component and water vapour the
other]). Provided they hold the conversion is exact.
ZTD is calculated as a sum of two integrals:
ZTD_hydrostatic = integral_of R_gas/g*k1*dp
ZWD = integral_of R_gas/(g*epsilon)*q*((k2-k1*epsilon)+k3/T)*dp,
ZTD = ZTD_hydrostatic+ZWD
with q being the specific humidity (density of water vapor relative to
total density of air), T the temperature, dp the pressure increment,
epsilon is the ratio of the molecular weight of water vapor relative to
the mean molecular weight of dry air (approximately 0.622), R_gas is the
gas constant for dry air (we use R_gas=287.04 J/(kg*K) ), g is the
net gravitational acceleration at the surface, assumed constant through
the part of the atmosphere modeled by us (we use g=9.80665m/s^2).
The IWV is calculated as an integral over,
q*dp/g.
For the constants we use:
k_1= 7.76 *10^-7 K/Pa
k_2= 7.04 *10^-7 K/Pa
k_3= 3.739*10^-3 K/Pa
(Bevis at al, 1994, Jour. Apl. Met., vol 33, p379)
Further details about the calculations can be found in
Vedel, Mogensen, and Huang, Phys. Chem. Earth, 2001, vol. 26, p. 497.
Format of data provided.
------------------------
The data are written into one file per station per day. The format of
each file is a follows:
content variables units
-------------------------------------------------------
station id name -
text line -
position lat, lon, alt(gps), alt(hirlam) deg, deg, m, m
time (starting) year month day hour yr,mth,day,min
# data points # records (lines) in body part of file
Each line of the body will have the following content:
------------------------------------------------------
hour, min, ztd_a, zwd_a, iwv_a, ps_a, ztd_h, zwd_h, iwv_h, ps_h, T2m_h, q_h
where hour and min specifies when the data are valid (in hours and minutes)
subscript a indicates the value is for the GPS altitude (antenna) level, while
subscript h indicates the value is for the HIRLAM surface.
ztd= zenith total delay in m
zwd= zenith wet delay, in m
iwv= integrated water vapor, in kg/m**2 (approx mm)
ps= surface pressure, in Pa
T2m= temperature (2 meter above ground), in K
q2m= specific humidity 2m above ground, in kg/kg
The fortran code used for the writing of the header is:
write(16,'(a5)')name(1:5)
write(16,'(a35)')'Data from NWP HIRLAM derived at DMI'
write(16,'(4f12.5)')lat,lon,gpsalt,hirlam_alt
write(16,'(I4,I3.2,I3.2,I3.2)')year,mon,day,minute
write(16,'(i4)')nrec
The fortran code used for the writing of the body data is:
write(16,1001)hour,minute,
& ztd_a,zwd_a,iwv_a,ps_a,ztd_h,zwd_h,
& iwv_h,ps_h,T2m_h,q2m_h
1001 format(i3,i3,3f10.4,f10.2,3f10.4,f10.2,f8.2,e11.4)
The names of the files will be yyyymmdd_cccc, where yyyymmdd specifies year,
month, and day, and cccc is the station id. For example data for station ZIMM
on 2003 May 6 will be in 20030506_ZIMM.
All files for one day are collected into one gzipped tar archive which is made
available on the DMI ftpserver. These data will become available also at
ACRI-ST, which is to operate a ftp-server for data in TOUGH. On the
DMI server the data will be stored as met/hirlam/yyyy/yyyymm/yyyymmdd.tar.gz
Each archive will remain on the DMI ftp-server for at least a month, longer
if space permits.
Updates are done daily. The extraction performed on a given day covers the
day before.
In the file 'gpslist.dat' providing the coordinates of all GPS sites for easy
reference each line contains:
name, latitude [deg], longitude [deg], altitude [m, geometric relative to geiod]
Access to DMIs ftp server.
------------------------
To connect to the DMI ftp-server, do a ftp to ftp.dmi.dk, specify tough
as userid and GPS2user as password. You'll be located
at the top TOUGH directory on the server upon being logged in. The ztd data
will be in met/hirlam/ and its subdirectories.
About HIRLAM model results versus radiosonde reports.
-----------------------------------------------------
For people not familiar with meteorology it is important to notice
the difference between the two types of validation data provided by the DMI.
The values of ZTD (etc.) produced from the HIRLAM model fields are NOT
measurements. They are numerical weather prediction model estimates of ZTD
(etc.) at the locations in question, and thus subject to the constraints set by
the finite resolution and by the quality of the HIRLAM model itself.
Contrary to this the radiosonde reports provided are real measurements, that do
not depend on a meteorological model.