A description of NetCDF format is presented on this page with a goal
of a most brief and popular introduction to the random web searcher that could
have very limited prior knowledge of the topic.
We are encouraging to check web pages of
NetCDF Introduction and Resources and The NetCDF Classic Format Specification to get academic answers.
Members of scientific and engineering communities are making a lot of
measurement from different instruments and could want to be able to exchange
them. The obvious problem that there could be various ways to present one set of
data that create a virtual analogy of human languages. Even more complicated as soon
there is several operation systems that move us forward to the analogy
with communications of different species that was even broad
discussed only in sci-fi literature yet.
One of the methods of solving this problem is an creation of universal
format of data. If it exists the problem of conversion of data
from one format to other could be tremendously simplified. Let´s
presume that we have N formats of data. To be able to transfer
date from one to another we should need (N-1)2 software
translators (note that software codes need to be written to translate from
format A to B is not the same as for the task to convert from
B to A). If we would have an universal format the task will be
extremely downsized because only 2*N converters will need to
be created for each operation system. We will need only converters
in and out of this universal format. The NetCDF format is an attempt to
create such Universal DATA Format or (network Common
Data Format).
Let´s presume for specificity sake that we a measuring an concentration
of different elements on the surface of some sample. It doesn´t
matter for geology goal, space exploration or technology improving.
The data that we could want to write down are information about instrument,
operator, time of measurement, notes about a sample and of course
data of concentrations themselves. One can differentiate short
notes about experimental conditions from data in general and
named them attribute. An example of attribute is "temperature=36.6
grad. C".
One can present it as plain text other to use statements like
""temperature=36.6" and "temperature units=C".
Second presentation has an advantage of universality but in general is
a matter of style.
It is possible to implement both styles in NetCDF.
The attribute could be globally related to all set of data in
the file or locally related to one set of data belongs to one variable.
The data itself could be presented as one or multidimensional arrays.
Technically all arrays in files are one-dimensional. To describe
structure of arrays of data a special member of the format - dimension
are included. For two-dimension array they are analogs of number of rows
and columns in spreadsheet table.
In NetCDF file at first there are declaration of all available dimensions
by name and size and then for each variable they are selected from the stock.
The variable can be considered as a set of data inside of all
collection of data presenting in file.
It can contain any number of local attributes and array of data
each dimensional structure described by dimensions.
To have additional description of organization of data in NetCDF file one can take a look on the page that described a transformation of data in XML format into NetCDF as soon XML format is reflected in readable form a content of NetCDF file.
Jun. 30, 2017; 10:02 EST