As meteorological ensembles (met-ensemble) become available as part of real-time operation at more and more NWP centers, it’s a growing area that many downstream prediction systems which strongly depend on meteorological forecasts as their inputs are also gradually deviating from traditional single-value input paradigm by actively testing how to couple with a met-ensemble system to quantify the forecast uncertainties in their downstream predictions.

Such downstream systems are but not limited to hydrology, air quality, transportation and dispersion, ocean waves, ice drifting, costal storm surge, and electricity generating etc.. Hydrological prediction is sensitive to precipitation information. Uncertainty in precipitation amount and type (liquid or solid) and maybe also in temperature will certainly cause large uncertainty in hydrological prediction of flooding, runoff and stream/river flow in both short term such as flash flood and long term such as snow-melting. Unreliable probability of precipitation forecast caused by model bias and imperfect met-ensemble as well as the mismatch in spatial scale between coarse model resolution of met-ensemble and fine scale of river basin or catchment are main challenges for a hydrological prediction system to correctly use met-ensemble information (Franz, et. al., 2005; Schaake, et. al., 2006). Therefore, downscaling and post-calibration of met-ensemble data is extremely important to hydrological application. Many meteorological fields such as temperature, advection, convection, precipitating process, radiation and especially surface and planetary boundary layer (PBL) properties like turbulence play important roles in controlling air quality and the transportation and dispersion process of pollutant. Those meteorological fields often exhibit large forecast uncertainties. Therefore, predicting air quality and dispersion process must suffer large uncertainty too. This issue has now been paid much attention by air-quality modelers and homeland security dispersion modeling community although how to properly simulate PBL-related uncertainties in a met-ensemble is still an issue yet to be researched. Since predictions in ocean surface wave, ice drifting and coastal storm surge are mainly driven by strong surface wind which possesses large forecast uncertainty, coupling with met-ensemble is also underway in those downstream prediction systems. To quantify such forecast uncertainty, NCEP has already implemented a wind-driven ocean-wave ensemble system in operation (Chen, 2006). Application to electricity generation is also popular in energy companies (Stensrud et. al., 2006). In all those downstream applications, most of them directly use individual met-ensemble members to drive a multiple of downstream predictions (a more expensive way in computing), while some use only forecast variance derived from met-ensemble to quantify uncertainty of a downstream prediction (a less expensive way in computing) such as in dispersion modeling (Warner et. al., 2002).