APPENDIX B: A BRIEF COMPARISON OF ECMWF, FNMOC, NMC, AND JMA GLOBAL NUMERICAL WEATHER PREDICTION SYSTEMS

1. Introduction

Numerical weather prediction (NWP) systems (NWPS) are used to determine the future state parameters of the atmosphere by numerical integration of hydrodynamic equations from an initial state. NWPS are cybernetic because they combine human and computer elements. The future of NWPS depends on computer power, observation data amount and data analysis methods, efficiencies of computational methods, understanding of physics, rigorous mathematic formulations of these physics, and interpretation of numerical weather prediction products. The computer program portion of a NWPS generally consists of four parts. They are:

(1) Data Analysis

The data analysis part of NWPS provides equally spaced data from unevenly spaced observations (reported data), climatological data and information, bogus data, and evenly spaced model output data. In some data analysis schemes, data nudging procedures are used to extract information from data taken at asynoptic times, such as, satellite sounding data, radar data, wind-profiler data, aircraft and ship report data, and surface observations.

Many automated data analysis methods have been developed because the ever increasing of data volume. The quality of an automated analysis method depends on the initial data sampling rates and data coverage, information content in the data, data error sources (such as instrument and observation, coding and formatting, transmission, or human errors), quality control methods, spatial and temporal coherence of data and data errors, and dynamic or thermodynamic consistency between data according to idealized states of the atmosphere.

Many operational data analysis procedures use NWP model output data as their first guess (for example, six-hour wind and thickness predictions), especially in regions where observations and climatological information are sparse. For forecasters, it is important to note that a model data analysis and interpretation are model dependent.

(2) NWP Model Initialization

The equally spaced data set obtained from a data analysis method may or may not be the initial data of a NWP model. The purpose of NWP model initialization is to remove NWP-model undesired gravity waves from the equally spaced data set. These gravity waves have several common characteristics. For example, they have small horizontal wavelength and vertical wavelength (small in terms of model horizontal grid distance and troposphere depth), short period (about 1 hour or less), and significant amplitude (about 1 hPa surface pressure). These gravity waves can be misrepresented (aliasing), or can grow artificially in a NWP model; thus, they are often labeled as noise in the equally spaced data set and should be removed. The method of model initialization is to eliminate the divergence due to a noise-like gravity wave (or wave mode) by adjusting the velocity and pressure fields associated with that wave. NWP model initialization techniques work extremely well for the dry portion of a model atmosphere that is located away from the equator. The resulting data set after NWP model initialization technique has been performed is the initial data set for the NWP model under consideration.

NWP model initialization is done either within the data analysis step or as a separate step which uses equally spaced data produced by the data analysis step. The choice of which type initialization to use is determined by local requirements.

(3) NWP Model

A NWP model is a computer program system that includes application, database, communication, interface, scheduling and system programs. This computer program system contains model physics for the earth's atmosphere, data flow representing data access and storage, as well as computational procedures. The NWP model is designed to predict atmospheric state parameters, such as the 500-hPa (or 500-mb) height and the 925-hPa (or 925-mb) winds, from a set of initial conditions and boundary conditions. The NWP model predicts grid-size variables such as geopotential height, winds, temperature and moisture distributions. The sub-grid size phenomena, such as local precipitation (i.e. rain, shower, snow or hail) amount, local winds, fog and visibility, are obtained by using empirical and diagnostic formulas (such as model output statistics techniques). There are local and regional numerical weather prediction models, or empirical methods which make use of these NWP model prognostic parameters.

(4) NWP Model Output

NWP model output data dissemination, display, discernment and interpretation are primarily driven by users' needs, and communication and display technologies. Possibilities include 1-, 2-, 3-, or 4- dimensional contour drawing, displaying and looping, stability indices and model output statistics (MOS) computations. MOS is a practical approach to provide statistically significant relationships between model output data and historical meteorological data. In this way the large-scale NWP model output can be extended for local weather forecasts, and local forecast accuracies can be enhanced. Model output interpretation is a talent that depends heavily on forecasters' experience and understanding of a NWP model and weather phenomena.

2. MODELING TRENDS

It is important to notice that almost all of the techniques adopted in a NWP system are evolving. Several clear trends in operational global modeling have been observed:

(1) Modular structure of model ingredients (including data, model physics, and computational methods).

(2) Centralization of global NWP systems (e.g., the consolidation of European modeling efforts at the European Center for Medium-range Weather Forecasts) to take advantage of advanced computation and communication power.

(3) Increasing global NWP models' spatial resolutions from "coarse" to "fine". A fine resolution model can represent more detailed structures embedded in a weather system than a coarse resolution model does. For example, TC structure will play an increasingly important role in global modeling as horizontal resolution approaches 50 km.

(4) Increasing research effort in data utilization, model completeness (in terms of physics formulations and phenomena inclusions as well as model resolutions), model output interpretation and computational methods.

(5) Automation of data disseminations. More effort will be spent for formulating, testing, verifying, and evaluating of these automatic or semiautomatic procedures.

(6) Increasing the forecast period from short-range, (up to 3 days), medium-range (3-7 days), long-range (1-2 weeks), monthly, seasonal to interannual.

(7) Development of ensemble forecast techniques. In general, there are two classes of ensemble forecast techniques; one is running the same NWP model with slightly different initial conditions, the other is running different NWP models including boundary conditions with the same initial conditions. These techniques are useful when the atmospheric behavior is closer to a chaotic state than a deterministic one.

The Table B1 (i.e., Tables B1-1, B1-2, B1-3, B1-4, B1-5, B1-6, B1-7, B1-8, and B1-9) compares four global NWP systems: ECMWF, U.S. FNMOC, U.S. NMC, JMA. The Table B2 (i.e., Tables B2-1, B2-2, B2-3, and B2-4) compares methods of initial specification of a TC for these global NWP systems.

References of Appendix B

The following references are helpful, but not cited.

Barker, E., 1992: Design of the Navy's multivariate optimum interpolation analysis system. Wea. Forecasting, 7, 220-231.

 

Bayler, G. and H. Lewit, 1992: The Navy operational global and regional atmospheric prediction systems at the Fleet Numerical Oceanography Center. Wea. Forecasting, 7, 273-279.

 

Bonner, W. D., 1989: NMC overview: Recent progress and future plans. Wea. Forecasting, 4, 275-285.

 

Clune, W. M., P. A. Harr, and L. R. Brody, 1991: The man- machine mix in operational product quality control and verification at Fleet Numerical Oceanography Center. Wea. Forecasting, 7, 280-287.

 

Derber, J. C., D. F. Parrish, and S. J. Lord, 1991: The new global operational analysis system at the National Meteorological Center. Wea. Forecasting, 6, 538-547.

 

ECMWF, 1988: The parameterisation of convection in the ECMWF forecasting model. ECMWF Newsletter, 43, 6-12

 

_____, 1989: Use of bogus observations to improve the analysis and forecast of tropical cyclones. ECMWF Newsletter, 45, 3-8.

 

_____, 1991: Development of the operational 31-level T213 version of the ECMWF forecast model. ECMWF Newsletter, 56, 7-13.

 

Goerss, J. S., and R. A. Jeffries, 1994: Assimilation of synthetic tropical cyclone observation into the Navy operational global atmospheric prediction system. Wea. Forecasting, 9, 557-576.

 

_____, and P. Phoebus, 1992: The Navy operational atmospheric analysis. Wea. Forecasting, 7, 232-249.

 

Hovermale, J. B., and R. E. Livezey, 1977: Three-year performance characteristics of the NMC hurricane model. Amer. Meteor. Soc., 11th Tech. Conf. on Hurricanes and Tropical Meteorology, Miami, Florida. 122-125.

 

JMA, 1993: Outline of operational numerical weather prediction at Japan Meteorological Agency (JMA). Appendix to progress report on numerical weather prediction. Available from Numerical Prediction Division, JMA. 128 pp.

 

Kalnay, E., M. Kanamitsu, and W. E. Baker, 1990: Global numerical weather prediction at the National Meteorological Center. Bull. Amer. Meteor. Soc., 71, 1410-1428. Kuo, H. L., 1965: On the formation and intensification of tropical cyclones through latent heat release by cumulus convection. J. Atmos. Sci., 22, 40-63.

 

Lacis, A. A., and J. E. Hansen, 1974: A parameterization for the absorption of solar radiation in the earth's atmosphere. J. Atmos. Sci., 31, 118-133.

 

Lord, S. J., 1991: A bogusing system for vortex circulations in the National Meteorological Center global forecast model. Amer. Met. Soc., 19th Conf. on Hurricanes and Tropical Meteorology, Miami, Florida. 328-330.

 

_____, 1993: Recent developments in tropical cyclone track forecasting with the NMC global analysis and forecast system. Amer. Met. Soc., 20th Conf. on Hurricanes and Tropical Meteorology, Houston, Texas. 290-291.

 

Mathur, M. B., 1988: Development of the NMC's high resolution hurricane model. Amer. Met. Soc., 17th Conf. on Hurricanes and Tropical Meteorology, Miami, Florida. 60-63.

 

Mesinger, F., Z. I. Janjic, S. Nickovic, D. Gavrilov and D. Deaven, 1988: The step-mountain coordinate: model description and performance for cases of Alpine lee cyclogenesis and for a case of Appalachian redevelopment. Mon. Wea. Rev., 116, 1493-1518.

 

Nelson, C. A., and W. T. Aldinger, An overview of Fleet Numerical Oceanography Center operations and products. Wea. Forecasting, 7, 204-219.

 

Onogi, Kazutoshi, 1993: JMA operational tropical cyclone bogusing method in global analysis. 1 pp. Available from the Japan Meteor. Agency.

Rennick, M. A., 1989: Moisture analysis system, final report. Martin Marietta Data Systems. 33 pp. Available from FNMOC, Monterey, CA. 93943.

 

Rosmond, T. E., 1992: The design and testing of the Navy operational global atmospheric prediction system. Wea. Forecasting, 7, 262-272.

 

Sato, N., 1990: Numerical weather prediction system in Japan Meteor. Agency. 18 pp. Available from the Japan Meteorological Agency.

 

_____, P. J. Sellers, D. A. Randall, E. K. Schneider, J. Shukla, J. L. Kinter III, T-T. Hou, and E. Albertazzi, 1989: Effects of implementing the simple biosphere model in a general circulation model. J. Atmos. Sci., 46, 2757-2782.

 

Sheets, R. C., 1990: The National Hurricane Center - past, present and future. Wea. Forecasting, 5, 185-238.

 

Trinh, V. T., and T. N. Krishnamurti, 1992: Vortex initialization for typhoon track prediction. Meteor. and Atmos. Phy., Vienna, Austria, 47, 117-126.

Appendix A Appendix C

Chapter 5