MODEL DESCRIPTION and CONFIGURATION
A. MESOSCALE MODEL VERSION 5
The forecasts on this website are based on computer runs using the MM5 mesoscale model. (http://www.mmm.ucar.edu/mm5/mm5-home.html) The MM5 is probably the most popular mesoscale weather model and is used in research worldwide. It was developed at the National Center of Atmospheric Research in Boulder Colorado, and at Pennsylvania State University. The MM5 is state of the art using the most up to date meteorological research for the model's formulation and set up. The PAGASA forecast uses the latest version V3.
The Domains
The PAGASA MM5 model run consists of two domains; the outer domain and the nested domain. Each domain covers a specified area that is run at particular horizontal and vertical resolutions within MM5.
The outer domain has a 60 km horizontal resolution. It is the largest domain, consisting of 91 by 91 grid points making an approximately 5500 by 5500 km square centered in the Philippines. The domain coverage is from 105oE-152oE and 5oS to 35oN. This domain is designed to simulate larger scale features that are affecting the Philippine region and adjacent areas, for example monsoons, tropical cyclones and other tropical weather disturbances.
Nested within the outer domain is a 20-km resolution domain (shaded area) consisting of 71 by 71 grid points that covers the Philippine islands. This domain is designed to simulate small scale features.
B. WAVE MODEL
Recent wave forecasting models are called Numerical/Spectral models because these evolved from the time-dependent energy balance equation of the wave spectrum. Such models treat the various processes that control ocean waves. (e.g. growth, propagation and dissipation) Numerical/spectral model requires considerable computational resources. So, that, further assumptions have to be employed to solve the balance equation. For this reason, the hierarchy of wave models arose to optimize computer capabilities. These are now referred to as; 1st, 2nd or 3rd generation wave models. The wave model developed by METEOFRANCE (called VAG Model, Guillaume, 1987) is adapted to forecast the waves over the Philippine waters (104-135°E, 2 - 24°N). The domain is divided into 125x89 grids spaced at 0.25°. VAG is a 2nd generation wave model applicable for deep water. It is used for operational purposes over the Atlantic Ocean and Mediterranean Sea.
INPUT/OUTPUT DATA
Winds from the Global Spectral Model are used to drive (input) the Philippine waters and obtain the wind-generated waves (Wave forecasts) for 72 hours (Output).
WAVE MODEL for Shallow Water
The Numerical Weather Prediction Group (NWPG) of PAGASA has started the implementation of a wave model to predict waves over shallow waters. It shall be implemented from one basin to another. Hence, the Philippine waters shall be divided into several basins.
C. HIGH RESOLUTION MODEL
The high resolution model, popularly known as HRM, is a regional numerical model used for short-term forecasts. It is developed by Detlev Majewski of the Deutsche Wetterdienst (DWD), the meteorological office of Germany. It is based on a set of equations that describe the evolution of variables such as temperature, wind speed, humidity, and pressure that define the weather or the state of the atmosphere.
The process starts with analyzed data using observations and previous forecast to obtain the best estimate of the current state of the atmosphere. A global model is ran to produce a forecast. This stage is done by the DWD, the meteorological office of Germany. The Philippine area of interest is then singled out to provide the initial and boundary conditions of the limited area model HRM and sent to PAGASA.
The HRM is hydrostatic and use a hybrid sigma pressure coordinate system. Near the surface, the model follows the terrain with sigma coordinates while the upper atmosphere employ the pressure coordinate system.
The physical parameterizations of HRM include a
At present, PAGASA use 14 km horizontal resolution (0.125° x 0.125°) with 301 x 201 grids and 32 vertical layers. The HRM is ran operationally twice a day: 0000Z and 1200Z, and is fully automated from downloading input files using the Preginet server to graphics display of outputs. With this configuration, the total processing time on 8 dual processor XeonEM64T, 512 MB memory, 2.4 mHz is about 70 minutes for a 78 hour forecast. Downloading the initial and boundary data files from the DWD website through Preginet starts at 11:00 AM and ends at around 12:15 noon. Model outputs are ready by 1:30 and are available from the PAGASA server for PAGASA forecasters to consider in their fomulation of their forecasts. Available outputs at the PAGASA server along the surface are rainfall, temperature, winds and pressure. At different levels of the atmosphere, the following variables are available: temperature, moisture, winds, geopotential heights.
The HRM is being used operationally by several countries all over the world. Among the users are the meteorological services of Brazil, Botswana, Bulgaria, Cyprus, Jordan, Israel, Italy, Kenya, Macedonia, Malawi, Mozambique, Oman, Romania, Spain, United Arab Emirates and Vietnam, University at Zurich, Guangzhou Regional Service of China, India Space Physics Laboratory.
The HRM is written in fortran 90 with some C subroutines for GRIB encoding/decoding. At PAGASA, parallelization is based on MPICH2 using 8 Xeon EM64T nodes. The Intel Fortran Compiler (IFC), non-commercial version, is used to compile the source program and Debian is the Linux operating system.
Using IFC for computers using Linux is one of the great advantages of HRM. Other numerical models such as MM5 require the use of Portland Fortran Compiler which requires a license to operate. Another big plus is the availability of input files earlier than other files such as the GSM or Global Spectral Model. This insures the timeliness of the forecast.
There is now an updated version of HRM of which the NMG will incorporate. Meanwhile, output display is continuously improved on for easy access, interpretation and better suited to the needs of the users. Future plans include using a higher resolution from 14 km to 7 or 2 km and data assimilation.
D. The Eta Model
The Eta model is a state-of-the art atmospheric model popularly known for its quasi-horizontal vertical coordinate. This model was initially developed in the 1970's in the former Yugoslavia (Mesinger & Janjic, 1974), was upgraded in the early 1980s (Janjic, 1984; Mesinger et al., 1988) and since then has undergone further developments notably at the NCEP (Janjic, 1990; Mesinger & Lobocki, 1991). Presently, the eta model in its various versions is being used worldwide for operational and research purposes.
The Eta model currently used by PAGASA is designed to simulate the weather conditions covering the entire Philippine archipelago and adjacent areas (101 deg E to 149 deg E and 6 deg S to 28 deg N) at a horizontal resolution of 28-km centered at 11 deg. N latitude and 125 deg. E longitude. It uses the output of the Japanese Global Spectral Model (GSM) for the models' initial atmospheric conditions.
E. The Storm Surge Model
This is an experimental storm surge model based on JMA Storm Surge Model, a numerical model developed by Japan Meteorological Agency (JMA) to simulate and predict storm surges mainly caused by tropical cyclones. The numerical scheme of the model is based on the shallow water equations and thus this model is two-dimensional.
The model has features listed below:
The model mainly uses Smith and Sandwell 2 minute bathymetry data distributed by University of California, San Diego, and ETOPO5 distributed by NGDC, NOAA.
Forecast stations covers the entire country. As a starting point, we utilize 36 stations covering Luzon, Visayas and Mindanao.