Modern operational methods of numerical weather prediction, such as “ensemble forecasting,” allow assessments of state-dependent predictability to be made. This means that forecast-specific estimates of the forecast standard errors are possible. Quantitative estimates of forecast uncertainty are often not communicated to the public as it is unclear what the value of this information will be to people who must make weather-dependent decisions. Using laboratory-based methods developed by experimental economists to study individual choice it is found that nonspecialists are able to make better decisions that increase their expected reward while reducing their exposure to risk, when provided with information about the day-to-day uncertainty associated with temperature forecasts. The experimental framework used herein may provide a useful tool for evaluating the effectiveness with which weather forecasts can be communicated to end users.

A climatology of nonfreezing drizzle is created using surface observations from 584 stations across the United States and Canada over the 15-yr period 1976-90. Drizzle falls 50-200 h a year in most locations in the eastern United States and Canada, whereas drizzle falls less than 50 h a year in the west, except for coastal Alaska and several western basins. The eastern and western halves of North America are separated by a strong gradient in drizzle frequency along roughly 100°W, as large as about an hour a year over 2 km. Forty percent of the stations have a drizzle maximum from November to January, whereas only 13% of stations have a drizzle maximum from June to August. Drizzle occurrence exhibits a seasonal migration from eastern Canada and the central portion of the Northwest Territories in summer, equatorward to most of the eastern United States and southeast Canada in early winter, to southeastern Texas and the eastern United States in late winter, and back north to eastern Canada in the spring. The diurnal hourly frequency of drizzle across the United States and Canada increases sharply from 0900 to 1200 UTC, followed by a steady decline from 1300 to 2300 UTC. Diurnal drizzle frequency is at a maximum in the early morning, in agreement with other studies.Drizzle occurs during a wide range of atmospheric conditions at the surface. Drizzle has occurred at sea level pressures below 960 hPa and above 1040 hPa. Most drizzle, however, occurs at higher than normal sea level pressure, with more than 64% occurring at a sea level pressure of 1015 hPa or higher. A third of all drizzle falls when the winds are from the northeast quadrant (360°-89°), suggesting that continental drizzle events tend to be found poleward of surface warm fronts and equatorward of cold-sector surface anticyclones. Two-thirds of all drizzle occurs with wind speeds of 2.0-6.9 m s Ϫ1 , with 7.6% in calm wind and 5% at wind speeds у 10 m s Ϫ1. Most drizzle (61%) occurs with visibilities between 1.5 and 5.0 km, with only about 20% occurring at visibilities less than 1.5 km.

No abstract