There are a number of reasons for this. For one, many days are benign and relatively easy to forecast. Fancy weather sensors don't help much in such cases. The real challenge comes when weather turns restless, when a storm is brewing or an extreme event is about to strike. In such times forecasts become less reliable — a small effect on overall accuracy statistics but potentially large in dollars and human toll. These costly events are the focus of sensor and forecasting research. Here we see a supercell storm from Nebraska in 2004. Photo by master storm chaser Mike Hollingshead.