Lessons Learned in Developing PERP (Python Environment for Radar Processing)

Our work at the National Center for Atmospheric Research includes processing data from research meteorological radars. We evaluate and tune signalprocessing algorithms that process as much as 1500 megabytes of data per hour in real-time. In the past year, we have designed and implemented the Python Environment for Radar Processing (PERP) using Numeric Python to aid in our research. Because Python is an interactive, interpreted language that does not require the compile and link steps needed by compiled languages, Numeric Python is a very productive programming environment. We were pleased to find that Numeric Python was also quite efficient in executing our algorithms. We found both SWIG and CXX quite valuable in writing our Numeric Python extensions. Although we did encounter some challenges in learning Numeric Python and debugging our extensions, Python allowed us to build a flexible, yet efficient, system to process our large data-sets and test our algorithms.