연세대학교 수치모델링연구실 연구원들이 APEC 기후센터(APCC)를 방문해 APCC 연구진들과 함께 협력 워크숍(Collaborative Workshop between APEC Climate Center and Numerical Modeling Laboratory)을 진행한다. 이는 2013년 1월 2일 오후 2시, 로터스국제회의장에서 진행되며 연세대학교 수치모델링연구실에서는 아래와 같은 제목과 내용으로 발표할 예정이다.
 

► Prof. Song-You Hong (Professor, Department of Atmospheric Sciences, Yonsei University)
Global Regional Integrated Model system (GRIMs)

 

► Mr. Myung-Seo Koo (Ph. D. Candidate, Department of Atmospheric Sciences, Yonsei University)
Development of an advanced global dynamical core with the stochastic tendency to improve weather and climate prediction
 

This study attempts to establish an advanced global spectral atmospheric model with state-of-the art numerical treatments and to investigate the effect of stochastic representation of model uncertainties on medium-range forecast for improving weather and climate predictability.

As an alternative to terrain-following sigma vertical coordinates that occurs pressure gradient error in upper layer of the atmosphere, a hybrid sigma-pressure vertical coordinates is newly devised and implemented into the double Fourier series dynamical core that is more computationally efficient in spectral representation than the traditional spherical harmonics dynamical core. In addition, the moisture transport scheme in the spectral models is replaced with a mass-conserving semi-Lagrangian scheme in order to avoid negative values arising from spectral transform in positively defined variables. In this process, a dimension-splitting algorithm is extended to three-dimensional advection, which is designed to preserve mass conservation on prescribed vertical coordinates. Parallel algorithm is additionally decomposed in the vertical direction to avoid communications overhead due to different decomposition method. The impacts of the implemented schemes on deterministic forecast skill are quantitatively assessed on various time scales.

To overcome a limited predictability due to nonlinear chaotic nature, the impact of stochastic representation of model uncertainty on medium-range forecast is also investigated, especially focusing on dynamical tendency for which stochastic approach has not been attempted in atmospheric numerical models. A stochastic perturbation is added to model tendency, with amplitude proportional to the tendencies itself. In addition, perturbation size is designed to be dependent on forecast time and vertical layer with three parameters: maximum interval, reference time, and temporal correlation. This stochastic forcing configuration is evaluated by deterministic verification in terms of large-scale diagnostics as well as precipitation. This study suggests that stochastic parameterization in both dynamical and physical tendencies would be useful for improving weather and climate prediction.

► Ms. Suryun Ham (Ph. D. Candidate, Department of Atmospheric Sciences, Yonsei University)
Development of an unified explicit cloudiness parameterization for general circulation model
 

To improve the simulation of cloud amount, several cloudiness parameterization have been proposed (e.g., Stephens 2005), which can be divided into the following three types. The first is based on the probability density function (PDF) of the subgrid-scale distributions of the cloud water and cloud amount (e.g., Smith 1990; Lohmann et al. 1999; Rotstayn et al. 2000; Tompkins 2002). In the second type, cloud amount is diagnosed from relative humidity (e.g., Slingo 1987; Slingo and Slingo 1991) or from both relative humidity and cloud water (e.g., Randall 1995; Xu and Randall 1996). In third type, cloud amount is predicted as a prediction variable using an equation considering the sources and sinks of cloud amount (e.g., Tiedtke 1993). These cloud parameterizations have been used in both operational NWP models and GCMs. Recently, from the evaluations large-scale feature and cloud-radiation feedback in the simulated climatology with the diagnostic and prognostic cloud schemes, it is evident that the diagnostic cloud generally tends to exhibit similar or better performance than the prognostic cloud experiments, even though the cloud microphysics and associated radiation and precipitation algorithms are more physically formulated (Shimpo et al. 2008; Hong et al. 2009). These studies suggest that the incorporation of sophisticated microphysics in GCMs requires not only an understanding of the cloud-radiation interaction for a specific microphysics scheme, but also formulation of a physically based cloudiness formula that considers the systematic biases of the large-scale features in the modeled atmosphere. In this study, to improve the simulation of weather and climate models, newly developed explicit cloudiness parameterization, which is included convective cloudiness, will be introduced. Also, interaction between the precipitation (convective, stratiform) processes and radiation algorithms will be investigated.

► Ms. Hyeyum Shin (Ph. D. Candidate, Department of Atmospheric Sciences, Yonsei University)
Analysis on resolved and parameterized vertical transport in convective boundary layers at gray-zone resolution
 

The gray zone of a physics process in numerical models is defined as the range of model resolution in which the process is partly resolved by model dynamics and partly parameterized. In this study, we examine the effects of grid size on resolved and parameterized vertical transport for horizontal grid scales including the gray zone. To assess how stability alters the dependency on grid size, four convective boundary layer (CBL)s with different surface heating and geostrophic winds are considered. For this purpose, reference data for grid-scale (GS) and subgrid-scale (SGS) fields are constructed for 50–4000 mesh sizes by filtering 25-m large-eddy simulations (LES) data.
As wind shear becomes stronger, turbulent kinetic energy and the vertical transport of potential temperature and momentum are more resolved for a given grid spacing. A passive scalar with bottom-up diffusion behaves in a similar fashion. For a top-down diffusion scalar, the cospectral peak scale of the scalar flux is larger than the horizontal size of the thermals and increases in time. For the scalar, the entrainment ratio, in conjunction with the shear, influences the mesh-size dependency of GS and SGS transport. The total vertical transport of heat and the bottom-up scalar is decomposed into a non-local mixing owing to the coherent structures and remaining local mixing. The contribution of the resolved parts is larger when roll-like structures are present than when only thermals exist, for both non-local and local fluxes. The grid-size dependency of the non-local flux and its sensitivity to stability predominantly determines the dependency of total (non-local plus local) transport.

* Contact: Ms. Sooyang Joo (051-745-3925)