WxMAP2—a python-based system for model diagnosis and weather forecast tool development

The concept of using the web to display "weather maps" or "wxmaps" for forecasting purposes goes back to the early 1990s and while the first version of WxMAP2 was developed for comparing climate models in the first Atmospheric Model Intercomparison Project (AMIP1) in 1995, a side application was displaying wxmaps from global numerical weather prediction models. This side application (WxMAP) was very popular with the general public and is now run operationally at several US Navy Meteorology and Oceanography centers (e.g., https://www.fnmoc.navy.mil/wxmap_cgi/index.html).

In 1999, the first WxMAP system was a Government Technology Leadership Award winner along with the AWIPS system of the National Weather Service. Although the WxMAP concept is well established, the underlying technology was a kludge of shell, perl, and GrADS (grads.iges.org) scripts and FORTRAN code, and the web was generated directly in html.

In 1997 I started the transition of the WxMAP system to python for python's cleaner syntax and maintainability; however, python was still used as procedural language in the same way as the original perl/grads/shell code. Last year I began the time-consuming task of reengineered the entire system, consisting of 1000s of scripts, written over 15 years, into python classes -- to take advantage of the python's object-oriented program (OOP) features and python interfaces to GrADS (opengrads.org) and other scientific packages including matplotlib and numpy/scipy.

In addition to the WxMAP2 model analysis and display applications, I also moved 30 years of tropical cyclone (TC) applications into the WxMAP2 python base for a unified system for handling both model and TC data. The paper will review the WxMAP2 python classes and demonstrate the benefit of OOP in an application that diagnoses the large-scale environmental factor the affect TC intensity.

WxMAP2 is now an open source project available at at wxmap2.sf.net.