HRM Model

National Center for HydroMeteorological Forecasting (NCHMF) has begun research on Numerical Weather Prediction (NWP) since early 1990. However, the progress has taken place only in 2000. The High-resolution Regional Model (HRM) originally developed by Deutcher WetterDienst (DWD) (for a shared memory computer, based on the OpenMP standard, firstly was installed in workstation IBM RS/6000 with 2CPU in 2000). Later, this code was developed for distributed memory systems like Linux PC clusters (based on the MPI standard) by the Vietnamese HRM group in co-operation with a local Institute of Mathematics, Vietnam Academy of Science and Technology (VAST). This new MPI version HRM had installed on a PC-based parallel computer with 8 processors and been running operationally since May 2002. This is the first NWP model running operationally in NCHMF and products from this model are good reference for daily forecast as well as inputs for the hydrological, wave and storm surge models.

Table 1 Short overview of the hydrostatic HRM

Prognostic variables	Diagnostic variables
- Surface pressure ps - Temperature T - Water vapour qv - Cloud water qc - Cloud ice qi - Ozone (optional) o3 - Horizontal wind u, v - Several surface/ soil parameters	- Vertical velocity ω - Geopotential φ - Cloud cover clc - Diffusion coefficients tkvm/h

Numerics of HRM

- Regular or rotated latitude/longitude grid - Mesh sizes between 0.25° and 0.05° (~ 28 to 6 km) - Arakawa C-grid, second order centered differencing - Hybrid vertical coordinate, 20 to 40 layers (Simmons and Burridge, 1981) - Split semi-implicit time stepping (Burridge, 1975); Δt = 150s at Δ = 0.25° - Helmholtz equation solved by a direct method (FFT and Gauss solver) - Lateral boundary formulation due to Davies (1976) - Radiative upper boundary condition as an option (Herzog, 1995) - Linear fourth-order horizontal diffusion, slope correction for temperature - Adiabatic implicit nonlinear normal mode initialization (INMI, Temperton, 1991) or diabatic (incremental) digital filter initialization (DFI, Lynch, 1997)

Physical parameterizations of HRM

- δ-two stream radiation scheme (Ritter and Geleyn, 1992) including long- and shortwave fluxes in the atmosphere and at the surface; full cloud - radiation feedback; diagnostic derivation of partial cloud cover (rel. hum. and convection) - Grid-scale precipitation scheme including parameterized cloud microphysics (Doms and Schättler, 2003) - Mass flux convection scheme (Tiedtke, 1989) differentiating between deep, shallow and midlevel convection - Level-2 scheme (Mellor and Yamada, 1974) of vertical diffusion in the atmosphere, similarity theory (Louis, 1979) at the surface - Seven-layer soil model including snow and interception storage (Heise and Schrodin, 2002)

Programming issues

- Coded in fixed format Fortran90; some C subroutines for GRIB encoding/decoding - Parallelization based on OpenMP for shared memory multi-processors and on MPI for distributed memory systems

Table 2. The parameters and products from HRM model ruuning at NCHMF

Initial & Bdy conditions	GME (DWD), 00 - 72h, every 3h
Horizontal resolution of GME	0.35° (~40 km)
Number of CPU	32 CPU
Integration domain	5°S-35°N, 80°-130°E (Big domain) 7.125°N-27.125°N, 97.125°-117.125°E (Small domain)
Horizontal resolution of HRM	0.25°, 0.125° (~ 28km & 14km)
Levels in vertical	40
Intergation step	120 (90) s
Initial model run	4 times (00, 06, 12, 18UTC)
Frequency for model outputs	Every 6 h
Standard levels for HRM outputs	surface and 4 levels (850, 700, 500,  200 hPa)
Alanyzed & forecasted fields	At surface: MSLP, Precipitation, 2m temperature and Wind at 10m At standard pressure levels: Vorticity, Divergence, Vertical velocity, Geopotenial and  Relative humidity