A Subkilometer-Gridlength Ensemble for Representing Stable Boundary Layer Forecast Uncertainty over Complex Terrain

Abstract

Abstract An ensemble prediction system featuring subkilometer horizontal grid spacing and high vertical resolution is used to quantify forecast uncertainty in the stable boundary layer (SBL). Diversity in initial conditions and/or planetary boundary layer/surface layer physics within the WRF Model provides ensembles with up to 12 members. WRF explicit ensemble data drive trajectory calculations and the Second-Order Closure Integrated Puff (SCIPUFF) model for hazard prediction. Explicit ensemble SCIPUFF forecasts are compared to single-member SCIPUFF forecasts leveraging WRF ensemble wind field uncertainty statistics. The performance of 1.3- and 0.4-km horizontal-gridlength ensemble configurations is evaluated for two case studies of differing flow regimes with respect to the Nittany Valley in central Pennsylvania where uncertainty in atmospheric transport and dispersion (ATD) is dependent on drainage flows and circulations related to trapped lee-wave activity. Results demonstrate that a 12-member ensemble provides reasonable spread in ATD forecasts. Additionally, single-member SCIPUFF surface-dosage probability forecasts using the meteorological ensemble statistics generally reflect the pattern while encompassing the hazard area given by the explicit SCIPUFF ensemble, but at a reduced computational cost. Low-level wind and temperature forecasts given by the 12-member, 0.4-km ensemble are improved significantly over corresponding 1.3-km ensemble forecasts. In general, the 12-member, subkilometer-gridlength ensemble configuration reliably captures temperature and wind fluctuations related to drainage flows and trapped lee-wave activity that directly impact ATD. Localized data assimilation positively impacts overall probabilistic forecast skill when trapped lee waves are present, and drainage flow case results appear more dependent on model physics than initialization strategy.