Understanding the Forecasting Process and Implications for Computing Tool Design

Weather forecasting is a complex intellectual task due to the complexity of atmospheric phenomena and the vast amount of data and tools used to measure, model, and present these phenomena. New interactive computing tools are continually being made available to meteorologists and existing tools are frequently upgraded. While operational meteorologists are often involved in the design of these new tools, no comprehensive task analysis of the forecasting process could be found in the literature. Understanding the overall forecasting process should be fundamental to a new system's User Interface Design and incorporation of new tools into the comprehensive operational meteorology environment. This study involved observations of forecasters at two NWS offices from which an overall task analysis was performed of the production of public forecasts. This task analysis revealed fundamental mental models used by forecasters: pattern abstraction, trend recognition, change detection, 4-dimensional data, time budgeting, forecast heuristics, data conversions, and forecast verification/feedback. Additionally, fundamental mental processing scripts were identified including: data sampling, communication/coordination, data relevancy filtering, and errant data elimination. This paper will provide an overview of the task analysis, mental models, mental processing scripts, and implications for future tools based on these observations.