With the background story that many students majored in mathematics or physics during the undergraduate study, I always feel necessary to write something for the beginners of oceanic sciences, aiming to improve their basic scientific skill on the specific research area.

It is not surprising that the atmospheric and oceanic sciences share some common features, for example, they both treat the earth-based four dimensional data-sets. In other word, longitude, latitude, depth, and time are the natural coordinates which we need to project the data into, and also where we extract data from.

We can imagine the difficulty during the data delivery without a common data format. As a result, the atmospheric and oceanic communities developed and shared an specified storage method, i.e. netcdf format.

I suggest my student abandon their window system, and move forward into the linux system. There are a variety of linux systems, and among them, the opensuse, ubuntu, redhat, fedora are the widely-used linux system among my surrounding colleagues.

We can not realize an scientific computing without some compilers of computational language. It is well-known that fortran language prevails in the earth science community. Fortunately, we can easily get the free distributed fortran compilers gfortran, or it is natural born in some linux systems. It is noted that gcc (compiler of C language) and g++ (compiler of C++ language) are the other two components besides gfortran in the  GNU compiler sets.

Once the GNU compilers were installed in your computer, you can install the netcdf in your system, with the guideline in the websites, as

http://www.unidata.ucar.edu/

and there are some examples of fortran/C/C++ scripts for reading or writing netcdf files:

http://www.unidata.ucar.edu/software/netcdf/examples/programs/

You also need write a Makefile for a project, which help the computer maintain and link the source files, see website:

http://www.chinaunix.net/old_jh/23/408225.html

Finally, we need some software help us visualization in the data post-processing, and I recommend the powerful and professional tool, NCL, which stands for the  NCAR Command Language. It is not easy to install the software from the source files, and I always download the binary directly, like

ncl_ncarg-6.2.0.Linux_RHEL6.2_x86_64_nodap_gcc472.tar.gz

For more information, visiting
http://www.ncl.ucar.edu/

Happy modelling.