Fundamental issues in numerical weather prediction

Weather-related disasters, unlike many other disasters, are often predictable up to several days in advance. This predictability leads to issues of preparedness that would depend directly on how much the predictions are trusted. This chapter gives insight into predicting the various types of weather disaster and how weather prediction products should be used by emergency managers who want to take into account the state of the science to determine how much they trust the forecasts. The chapter also explains the areas of current uncertainty in weather prediction, which are primarily in the areas of model physics and initialization data. Introduction A key tool in current-day weather forecasting is the numerical weather prediction (NWP) model, and it is usually the primary guidance for forecasts ranging from a few hours to a few days. NWP covers a vast span of temporal and spatial scales, and model domains range from global to continental to national to local. In this chapter, the range of scientific and technical issues that form fundamental limits to model predictions are addressed, with particular attention to the way these problems greatly depend on the particular scales of application required for the forecasts. Therefore, this chapter is organized somewhat according to the type of application, which is largely a function of the required forecast range.