Choosing a Method of Database Access

There are three main ways of accessing data from the MCD which have been implemented to date.

Firstly, a complete interface, MCDGM, is supplied with the Mars Climate Database as a set of FORTRAN source code in the mcd/mcdgm subdirectory. The MCDGM interface performs in a very similar way to MarsGRAM version 3.5. It is intended to make the database as easy to use as possible for those with prior experience of MarsGRAM as well as providing straightforward access for all users to the complete database, including the variability model and orographic wave model. It can be run in interactive or batch mode.

Secondly, it is possible to access the database directly from within any program, written in FORTRAN or C, by using the DRS library. This gives the most flexibility for particular applications, although it does demand a greater understanding of how the database and variability models, if they are required, should be used. Some examples of programs which open and read the database files are given, along with some useful general purpose routines, and the MCDGM interface itself also provides an example of source code which accesses the data files using the DRS library.

Thirdly, control files, instructions and example scripts (in the mcd/grads subdirectory), are provided for accessing the database using GrADS. GrADS is a freely available package for access, manipulation and display of earth science data which runs on many computing platforms. This provides a very easy method of examining and plotting mean and standard deviation data from the MCD in a variety of formats. It is not straightforward at present to use the variability model or orographic wave model from within a GrADS script; though GrADS or any other visualization package could, of course, be applied to the output from either of the first two access methods when these features are required. A second disadvantage is that GrADS requires a uniform vertical grid. Since the database has been stored in terrain-following $\sigma$ levels ($\sigma = p/p_s$, where p is the pressure and p_s is the surface pressure which varies as a function of position and time) in order to retain the model's high resolution near the surface, this means that it is not possible for GrADS to display data on a true height coordinate without writing external routines to read the entire database, convert to height coordinates by integrating the hydrostatic equation for each profile, add the height of the local surface above the geoid, interpolate onto a uniform height coordinate grid and then re-write the database and control files in this new form. This process is perfectly feasible using the FORTRAN routines supplied with the MCD, but requires disk space to store the new data and results in losing nearly all the high-vertical resolution, near-surface information. This is not worthwhile for simple data visualization and so the GrADS routines supplied access the standard $\sigma$ level database; $\ln (\sigma)$ is approximately proportional to height above the local surface for an isothermal atmosphere (very roughly for Mars, $z \approx 10\ln (\sigma)$km) and so a $\ln (\sigma)$ axis is adequate for most plots. For individual profiles, the surface pressure at that location and time can be read and multiplied by $\sigma$ to give a pressure-coordinate axis. If accurate height is a priority then access to the database should be made through either of the first two methods. Nevertheless, GrADS is recommended as a way of producing reasonable quality graphical output quickly and is ideal for examining one, two or three dimensional ``slices'' through the data.