MM5IDL: A Flexible Framework for Post—Processing MM5 Data

MM5IDL: A Flexible Framework for Post-Processing MM5 Data

A common approach for post-processing data from the MM5 weather model has been to convert the data from the native format to a more convenient one, i.e. netCDF, RIP, GRADS, etc. This approach has serious flaws if the dataset is large (waste of disk space) and/or extends over a long time period (some of the post-processing tools used for MM5 are limited to relatively short time-series).

In order to handle these shortcomings a novel software framework, the mm5idl, has been created. The package is targeted at metorological researchers using the MM5 model, and wanting to unleash the flexibility of IDL to analyse their simulation results.

At the core of the package is an IDL class hierarchy which deals with reading, and to some extent writing, files from the MM5 weather modeling system. On top of this core, a flexible framework for analysing the data has been built. The aim of the framework design is to enable the casual IDL user to add his/her own data analyses and mapping routines as "drop-in" classes. Programmatically, such analysis routines are then applied to time periods as simple, single-line constructs.

The framework is particularly well suited to batch processing of results from climatological experiments or other MM5 datasets with a large number of timesteps. Provisions are made to take care of broken output files (typically caused by model crashes), patching together datafiles across model restarts, and selecting specific periods for analysis. Data dependent conditions can also be provided to apply the post-processing under specific circumstances only.

At present, a number of post-processing tools have been developed under this framework. Data from 15 years of climatological downscaling experiments for Iceland is routinely processed with these tools. Tasks include data extraction for comparison with observations at meteorological stations, retrieval of subdatasets for inclusion into other models, and timeseries analysis. The MM5IDL package also forms the core of the post-processing software used for displaying daily forecasts, made for Iceland and surrounding waters, on the web (http://www.os.is/~or/vedurspa). Using the GUI capabilities of IDL, it is also straightforward to add a user interface layer to create a full-blown application.

The software is distributed under the GPL license.