Workshop: Observational campaigns for better weather forecasts

Often it is thought that observations exist to serve models. A commonplace idea in the atmospheric and climate sciences is that observations exist to 'verify', 'evaluate' or 'improve' models. In this talk I instead focus on how models can be used to guide observations. By identifying specific processes underlying a particular type of modeled behavior, models can instead be used to craft hypotheses that observations can be used to test. This is an old-fashioned an familiar idea for much of the physical sciences, but poorly practiced in atmospheric and climate science. Examples will be provided of where this has been successful, and why it might be useful to think of models as tools for guiding and sharpening observations, rather than the other way around.