NOMADS and using the DODS server HOWTO
Forecast System (GFS) models ability to forecast precipitation at
medium ranges is limited even if global ensemble forecasts are used,
but lets use the 2nd week forecast of precipitation from
the low resolution GFS (MRF) as an example in this demonstration.
Also below, we will use the date 20050315 as the example for the date
template YYYYMMDD. NOMADS, the web site you are viewing, is a
prototype, meaning not (yet) operational so the server(s) may not be
operating all the time. Having said all that, you can access the
precipitation from the GFS forecasts using our client examples (plot
(pdisp or “Great Displays”), ftp2u, and http
(Fast ftp) available on the web
http://nomad3.ncep.noaa.gov/ncep_data (can also change the 3 to a 5 for some backup)
There are many data set
entries but lets confine ourselves to the GFS low resolution (2.5
degree) data set. The other datasets work the same way. Begin by
clicking on a button on the table entry "GFS Low Resolution" about 3 lines up from the bottom of the
table. (The GFS reduces its resolution to 2.5X2.5 degrees after the 180-hr forecast, so to make a seamless data set from initial condition out to a 2 week forecast, requires that Nomads interpolate the forecasts before 180-hrs to 2.5 degrees. We have done this in the data set “GFS Low Resolution”. The table entries of “GFS High Resolution” contain data sets that are on a 1x1 degree lon/lat grid out to 71/2 day forecasts. ½ degree is coming soon!)
Click on one of the
following depending on what you need:
<plot>: to make a plot over an area in question of user selected variable(s),
<ftp2u>: to get the digital data "sliced and diced", then repackaged as a packed GRIB file over your area subset with the variables you want and sent to your ftp server, or ours, for web and web-ftp access.
<http>: "fast ftp" see http://www.cpc.ncep.noaa.gov/products/wesley/fast_downloading_grib.html
for fastest transmission for data sets that do not need an area subset.
<DODS>: Make our disk, your local disk! If you have a client like GrADS, IDL, or MATLIB or if you do not have such a client, but access to a web non-interactive download program like wget or wwwgrab (linux). Or if you have a common browser like Internet
Explorer, then start by getting the URL name of the DODS data set (GDS/DODS includes a meta-data file description in ascii text that describes each data set):
Click on the <DODS> button and follow the prompts to the "info" web page (for today) at:
which is called the meta-data description file and get the data location URL from the first
"DODS URL" line at the top of the description:
Paste this URL location to your empty URL-location in a web browser like
in Internet Explorer. For our example, use the 2 dimensional variable “surface total precipitation” [kg/m2], "apcp"
(a list of variables, units, etc is on the info page mentioned above and see also "prate"), add to the URL the characters
the query” to the variable apcp precip, beginning
from week 1 through 2 -- from the 7 day (168hr) to the 14 day (336hr) forecast on 12 hr output files [14:28]. If one wanted only forecasts at 00Z one could introduce a “stride”, a skip, [14:2:28].
At, for example, Abu Dhabi (lat=24.4,lon=54.7) on a 2.5 degree grid so [46:46][22:22] where latitude 46 is the number of 2.5 degree units from the south pole and longitude is the number of 2.5 degree units from Greenwich.
Putting it all together with no spaces or carriage returns
returns the forecast apcp at Abu Dhabi from day 7-14 on 12-hr intervals!
Place this URL in a suitable unix script as the quoted object of a wget (non-interactive web download)
and change the date and place it in cron for an automatic daily download of the information you need --
but please place a "sleep 30" (wait 30 seconds) if you place in inside a unix script for loop in case
of a run away script.
Try the DODS constrained query maker using global ensembles to find probabilities you select at:
http://nomad5.ncep.noaa.gov/cgi-bin/var/ensprob1.pl (also try ensprob2.pl). click yes to show the
URL's that are constructed by the program.