How the National Weather Service can implement, in a timely and continual manner, the rapid technological advances in the computing and software arena. T oday's weather forecasting services represent an impressive array of technology blended synergistically with scientific knowledge. It was less than a century ago that the possibility of numerically predicting the weather was suggested by Wilhelm Bjerknes (1904), Exner (1908), and by Cleveland Abbey (Gedzelman 1994). In 1922, Richardson attempted his ambitious and famous numerical forecast that failed. The reasons for the failure were many, but most significant was the lack of adequate observations, a condition that prevailed for more than 20 years (Lynch 1999). This changed when Motchanoff in Russia, Duckert in Germany, and Vaisala in Finland, independently developed the first practical radiosonde (Lange 1935). These elegant devices made feasible and affordable the measurements necessary to initialize numerical models. Still, the dream of numerical weather prediction remained elu