n fŻÈwêU n ÈwçtÁÊuï ú: 2011N617ú(à) 16:00`18:00 ê: sċww1Ù5K563ş yPz uPèĵF The Experimental Regional Seasonal Ensemble Forecast at NCEP uÒPF Dr. Hann-Ming Henry Juang (Environmental Modeling Center, NOAA/NCEP) uPv|F For the past years, the experimental regional seasonal ensemble forecast (RSEF) at NCEP by using regional spectral model has been conducted under the support from NOAA funds. The main idea is to downscale from the NCEP CFS (Couple climate forecast system) over USA to examine the capability of the regional ensemble forecast for seasonal predictions, at the same time to provide downscaling data for fire danger seasonal prediction. Due to the resources limitation, a reduced set of model run with limited ensemble size and domain coverage only over CONUS is performed once in a month since 2003. Using this data set from this single model RSEF, including hindcast since 1982 up to now, we can have preliminary idea of possible capability and/or predictability of RSEF over CONUS. The statistical results from this long-term data set will be presented. The predictabilities of season, initial condition, diurnal cycle and different regions over CONUS will be investigated. In addition to present the results from the experimental monthly RSEF, several sensitivities of limited sets of RSEF have performed to examine the predictabilities of single model RSEF, such as boundary conditions, bias correction, and ensemble sizes. Finally, all other possibilities of configuration to improve RSEF will be discussed. Some results from most current project will be presented as well. yQz uQèĵF Advection of smooth and non-smooth tracers with spectral bicubic interpolation uÒQF Dr. Takeshi Enomoto (Disaster Prevention Research Institute, Kyoto U.) uQv|F Spectral bicubic interpolation (Enomoto 2008) is a simple and accurate scheme that uses the derivatives calculated from spectral coefficients in bicubic interpolation. It is intended for applications in spectral models traditionally used in general circulation models for numerical weather prediction (NWP) and climate simulations. In such models the derivatives are trivially obtained. The longitudinal, latitudinal and cross derivatives are used in addition to the grid point values at the four corners of the cell. Since bicubic interpolation is local, no halo points are required for interpolation except for grid points beyond the poles. Dissipation, inherent in semi-Lagrangian advection, is very small with this scheme. Enomoto (2008) has shown excellent accuracy of the scheme for smooth tracers by performing the test of Ritchie (1987). In this study the scheme is validated by the challenging tests that include non-smooth fields (Lauritzen and Skamarock 2010). The proposed scheme is compared with the quasi-cubic interpolation, which is a combination of one-dimensional linear and cubic interpolation (Ritchie et al. 1995) commonly used in NWP. In general spectral bicubic interpolation produces smaller error than quasi cubic interpolation. For example, the edge of the slotted cylinders is sharper and overshoots around the correlated cosine bells are reduced. The selective quasi-monotone filter (Sun et al. 1996) effectively removes most of short-wave noise. The fixer of Sun and Sun (2004) is used to conserve global mass. The filter and fixer degrade the accuracy very little and require negligible computational cost. References : Enomoto, T., 2008: SOLA, 4, 5--8, doi: 10.2151/sola.2008-002. Lauritzen, P. H. and W. Skamarock, 2010: available from http://www.cgd.ucar.edu/cms/pel/tracer-workshop.html Ritchie, H., 1987: MWR, 115, 608--619. Ritchie, H. et al., 1995: MWR, 123, 489--514. Sun W.-Y. and Sun M.-T., 2004: MWR, 132, 975--984 Sun W.-Y. et al., 1996: QJRMS, 122, 1211--1226. â˘íıĉF ]c Ĵj