You did try to use Mirador for subsetting and choosing the parameter of your choice? 

Yes.. Mr. Marcel, I have downloaded the said data from Mirador only, there are 8 files for each day (as it is a 3-hourly dataset).

My question is how to extract a complete annual time series as a single file ?

Hi Deepak,

Did you try this one then?

http://disc.sci.gsfc.nasa.gov/recipes/?q=recipes/How-to-Obtain-Spatially-Subsetted-Time-Series-Data-in-One-NetCDF-File-via-GDS

Dear Marcel,

Thanks for sharing the link, I tried to do the same thing after installing NCO libraries, but have no idea how to select any particular year.

Could you tell me how to specify the Year in the command, as there is no mention of the specific year of study, please see below:

=======================================================

The following command and options will download daily data over the eastern U.S. for Jun to Aug 2000:

ncks -O -v r -d time,151,242 -d lon,275.0,294.0 -d lat,25.0,40.0 http://disc2.nascom.nasa.gov:80/dods/3B42_V7_daily outfile_trmm.nc

where;

-d time,151,242 : time index for Jun 1 and Aug 31 2000. Note that time index starts as 0. The time index of a day is the total number of days from the begin day (Jan 1 2000, in this example) to this day.For example, the time index of Jun 1 2000 is 151 (=31+28+31+30+31).

-d lon,275.0,294.0 :longitude from 275.0 to 294.0 degrees

-d lat,25.0,40.0: latitude from 25.0 to 40.0 degrees

How is it selecting the year 2000 ?

_ _

Thanks & Regards

Hi Deepak,

Though I have no experience with this application, I guess there is a mistake in this example. The year 2000 is mentioned, but the series start at 1998 and the example selects data from June 1 until August 31 1998. I reckon that one should count the days from January 1 1998. So for selecting the year 2000 one should give as starting day 730 and ending the year with day 1095. 

Hi,

Without being an expert in the precise data-set from which you are willing to extract a time-series (I assume a single point time series), I guess the answer to your question is no, there is no easy way. You will have to work out the time-series yourself. This is the rule almost in all cases: data is free in the internet; you have to processes it yourself for your particular uses.

Below you will find an script that makes the work (for a single year) you are trying to do in a “dirty” way (with dirty I mean you will have to download a lot of data to retain just a small piece). Roughly talking, it will work as long as you have a machine with a common linux distribution (it is a bash script) an CDO (https://code.zmaw.de/projects/cdo) installed with compatibility for grib data (the data format you are willing to processes). The result will be netcdf file called “mytimeseries.nc” (to avoid this grib to netcdf format change remove the “-f nc” option from the cdo command within the loop). If that is not the case, it will illustrate the kind of steps you will need to follow to have the work done.

Note: I have chosen parameter 111.1 from the grib file, which (without going too deep in the documentation) seems to be the short wave radiation (and so I named it). I will have to work out the parameter IDs and related variables (I mean those of your interest) from the documentation.

Good luck!!

#!/bin/bash

#clean folder: remove grib and netcdf files

rm -f *.nc *.grb

data_source='ftp://hydro1.sci.gsfc.nasa.gov/data/s4pa/GLDAS_V1/GLDAS_NOAH025SUBP_3H'

#Select YEAR here

year='2001'

#Select location. Example: over Greenland (i=550,j=550) or (317.375E, 77.375N)

#Select LONGITUDE grid index (position of point in grid)

#You need to work out this for your area

#Equal idx1 and idx2 for single point time series, add range for area

idx1='550'

idx2='550'

#Idem for LATITUDE

idy1='550'

idy2='550'

for ((x=1; x

Thanks for the help Ganix.

I am new to this shell script thing. But, I tried to run your code as it is, except setting the path for cdo libraries.

Despite of several attempt it failed to download the data because of the error in the server response. I will appreciate if you can have a look in the script and the error mentioned below.

=================================================

Script

=================================================

#!/bin/bash

#clean folder: remove grib and netcdf files

rm -f *.nc *.grb

data_source='ftp://hydro1.sci.gsfc.nasa.gov/data/s4pa/GLDAS_V1/GLDAS_NOAH025SUBP_3H/'

#Select YEAR here

year='2001'

#Select location. Example: over Greenland (i=550,j=550) or (317.375E, 77.375N)

#Select LONGITUDE grid index (position of point in grid)

#You need to work out this for your area

#Equal idx1 and idx2 for single point time series, add range for area

idx1='550'

idx2='550'

#Idem for LATITUDE

idy1='550'

idy2='550'

for ((x=1; x ‘.listing’

Connecting to 10.3.100.207:8080... connected.

Logging in as anonymous ...

Error in server response, closing control connection.

Retrying.

===================================================

Thanks & Regards

Hi Deepak,

There might be another way to access the desired data. Starting at the link you gave, switch to the Giovanni column and choose your file. Then, zoom in for the region you choose or fill in the coordinates of the area to be studied. Keep it a bit small, otherwise the server will take a lot of calculating time in the preprocessor. Make the choice for your parameter and select time series under "Select Visualisation". After processing you get a graph, but the data can be downloaded too. You can download the data as batch in HDF, CDF or ASC format. Even if you select one parameter, the files can be large still.