An Intelligent User Interface to Support Weather Report Generation

Abstract

Modern-day computers have greatly advanced our ability to generate reasonably accurate weather forecast models in a timely manner. High-powered workstations have put the output from these models, as well as a tremendous amount of raw and analyzed data, at forecasters' fingertips. In fact, the sheer volume of available data can be overwhelming, making it difficult for forecasters to conduct detailed analyses of all the products available to them. Instead, they rely heavily on domain experts' distilled analyses of computer models. Though these computer models are the result of a far more extensive analysis of specific weather parameters than a forecaster could perform in real time, they cannot be relied upon to do the work of the forecaster?to produce and disseminate an accurate terminal aerodrome forecast (TAF) every 6 h, then monitor it for correctness and accuracy, and amend as necessary. Given the broad spectrum of variables interacting in the atmosphere, a forecast ?solution? to any given meteorological circumstance cannot be algorithmically found. Therefore, though it is advantageous to harness data storage and computational power to generate quality weather products, human reasoning is still a necessary part of the forecasting process. This article discusses the practical application of an object-oriented, knowledge-based ?critic? advice system to aid in the generation and monitoring of weather products. This system would provide domain- and situation-specific expert advice, throughout the life cycle of the product, to ensure weather product quality and improve accuracy in the reports meteorologists create.