Recipe: Matlab program as OPeNDAP client

Back to G5NR Data Access Guide.

Problem

By accessing the collection inst01hr_3d_T_Cv via the OPeNDAP server

http://opendap.nccs.nasa.gov/dods/OSSE/G5NR/Ganymed/7km

we want to read the surface temperature data inside the box bound by latitudes 25^oN, 50^oN and longitudes -130^oW, -65^oW for 2006/Sep/18, 9z and compute its min/max.

Solution

MATLAB version 2012a and later has native openDAP support. The following code has been tested with MATLAB version 2014a.

The metadata for the collection inst01hr_3d_T_Cv is available at

http://opendap.nccs.nasa.gov/dods/OSSE/G5NR/Ganymed/7km/0.5000_deg/inst/inst01hr_3d_T_Cv.info

Code

This code accesses the collection inst01hr_3d_T_Cv from the OPeNDAP server and reads a subset of the temperature data (all levels inside the bounding box specified above) and computes its max/min. It then computes the max/min of the above data at the surface (level=72).

NOTE:

1. Instead of reading a downloaded NetCDF-4 file, we read an OPeNDAP URL.
2. While in the downloaded file, the temperature variable appears in the uppercase (T), on the OPeNDAP server, this variable is in lowercase.
3. Via the OPeNDAP URL, we now have access to all times for which data exists. The hourly inst files are available starting at 2005/15/15, 2200z. Our desired time, 2006/09/18, 0900z is then the 11772^th step.

% opendap url
opdurl = 'http://opendap.nccs.nasa.gov:9090/dods/OSSE/G5NR/Ganymed/7km/0.5000_deg/inst/inst01hr_3d_T_Cv'

% bounding box
%   lons = -130:0.5:-65
%   lats =   25:0.5:50
imin = round((-130. + 180.)/0.5);
imax = round(( -65. + 180.)/0.5);
jmin = round((  25 +  90.)/0.5);
jmax = round((  50 +  90.)/0.5);
% corresponding array sizes
im = imax-imin+1;
jm = jmax-jmin+1;
lm = 72; % read all 72 levels

% read temperature inside the bounding box
% time 11772 corresponds to 2006/09/18, 9z
start = [imin, jmin, 1, 11772];
count = [im, jm, lm, 1];
fprintf('Reading Tsub (subset of T)...');
Tsub = ncread(opdurl, 't', start, count);
fprintf('done.\n')

% compute max/min of Tsub
fprintf('max(Tsub): %f\n', max(Tsub(:)));
fprintf('min(Tsub): %f\n', min(Tsub(:)));

% compute max/min of Tsub at surface
level = 72; % surface
Tsub_surface = Tsub(:,:,72);
fprintf('max(Tsub at surface): %f\n', max(Tsub_surface(:)));
fprintf('min(Tsub at surface): %f\n', min(Tsub_surface(:)));

Compile and link

We use the utility nf-config (included in the NetCDF-4 installation) to identify the linking rules

For a typical NetCDF-4 installation, the above code, g5nr_reader_dap.f90 can be compiled and linked to the NetCDF-4 library via

gfortran -o g5nr_reader_dap.x `nf-config --fflags` g5nr_reader_dap.f90 `nf-config --flibs`

creating the executable g5nr_reader_dap.x.

NOTE:

1. You can use your favorite Fortran compiler instead of gfortran.
2. If the NetCDF-4 library was built with parallel I/O support, you will need to use mpif90 to link, even if your code does not use the MPI library.

Run

Running the executable,

./g5nr_reader_dap.x

produces the output

Reading T (subset)...
T (subset):    305.6048       191.6956

See Also

1. File Spec: File:G5NR-Ganymed-7km FileSpec-ON6-V1.0.pdf
2. Recipe: Fortran program to read data from downloaded file