COUNTRY: Australia
CENTRE: NMOC Melbourne
January 2000

Ongoing throughout 1999: Testing and making systems Y2000 compliant.

17 June: More weight given to mean sea level pressure PAOBS in GASP.

27 July: Australian region Limited Area Prediction System (LAPS) upgraded to 0.375⁰ in the horizontal and 29 levels in the vertical.

3 August: Assimilation of ERS-2 altimeter significant wave height data into seastate system.

22 September: Tropical region Limited Area Prediction System (TLAPS) upgraded to 0.375⁰ in the horizontal and 29 levels in the vertical.

16 November: Replacement of the SE and SW MESO_LAPS by a single domain MESO_LAPS_PT125 system (0.125⁰ in the horizontal and 29 levels in the vertical).

25 November: Introduction of the 7-day Model Output Forecast (MOF) using GASP.

7 December: Introduction of TCLAPS for tropical cyclone prediction, featuring a relocatable domain.

The National Meteorological Operations Centre (NMOC) Melbourne's computing backbone is provided by 11 Hewlett-Packard 9000 series Unix servers, a NEC supercomputer and a Cray J90. Currently, 7 of the servers have 2 cpus and the remaining 4 have a single cpu. All run under HP-UX B.10.20.B with clock speeds of 100 Mhz and up to 1536MB of RAM. Disc storage of approximately 780 GB is associated with the servers. The NEC SX-4 supercomputer - which is shared equally with CSIRO - has a configuration of 32 processors, 8 GB of Main Memory, 16 GB of Extended Memory and 600 GB of disc, and runs under the SUPER-UX Unix 9.1 Rev 1 SX-4 Operating System. A Cray J90 (4 processors, Unicos 10.0.0.6, 128 MW memory, 64.0 GB), is also available for development work. Real time scheduling is currently carried out on one of the HP Unix servers. A FDDI link provides the operational connections between the Unix servers, the NEC and the Cray. A 1TB StorageTek Mass Store 4400 ACS (Automatic Cartridge System) provides the archive facility for ReelLibrarian, and 2 StorageTek 9710 Library Storage Modules (with a 34TB capacity) using 12 DLT, ie Digital Linear Tape, units provides the facility for Sam-fs. Peripheral equipment includes HP Design Jet 650C plotters, HP LaserJet 5 SiMX laser printers and an HP Deskjet 1600CM coloured printer. There are a number of LANs which provide access to internal users. A PC-based digital facsimile system (DIFACS) is used for product dissemination nationally.

Basically, the main numerically intensive analysis and forecast components of the NWP systems run on the NEC SX-4, while the less intensive tasks (for eg message recognition and decoding, data base ingestion and extraction, post-processing, display, etc.) tend to be carried out on the HP computers.

The overall computer environment in the NMOC is mainly Unix. A real-time data base, currently using ORACLE 7.3.3.4.0, is used for storage of observational data and grids from the various NWP systems. The U.S. Navy's Environmental Operational Nowcasting System (NEONS) software is commonly used for accessing the data base. The operational NWP models are written mainly in Fortran, with many of the associated files having the NetCDF structure. Most displays are produced using the NCAR graphics package and IDL. The CIM CAD package is used in the production of significant weather prognoses. Sam-fs and ReelLibrarian systems are currently being used for magnetic cartridge archives in the NMOC.

The DIFACS system is used to to disseminate a selection of basic analysis and prognostic charts, and some satellite imagery, to the Bureau's regional offices and some outside users. MCIDAS is used for comprehensive interaction and display of satellite imagery and products, observational and gridded data and is also a major component of the Australian Integrated Forecast System (AIFS). Products from the NWP systems are written to internal and external (www.bom.gov.au) web servers. Magnetic cartridge archives are kept of various NWP products with Australian region analyses available back to 1970 and Southern Hemisphere analyses back to 1972. Hard copy and microfilm archives of charts also exist. An aviation system, which interacts with the WAFS data, is used to view and display the data and prepare the various flight and route forecasts. Regular statistics (including S₁ skill scores, root mean square errors and anomaly correlations) monitoring and comparing the performance of the NMOC's NWP systems (and also some overseas NWP models) are also produced.

The following table gives a list of the major report types used in the NMOC Melbourne:

The following table gives approximate numbers of reports being received for a 24 hour period:

Automated. (Some manual intervention is available for correction of reports.) The observational data, along with NWP gridded data, is stored in a real-time relational data base system (ORACLE/NEONS).

Validity checks are currently confined to within the respective assimilation or analysis schemes. Some gross checking outside these schemes may eventually be installed.

Monitoring of the observing system is carried out. The quantity of data available is monitored in real-time to ensure that reports are being received reliably and are made available to the operational systems, both automatic and manual. For the global system, statistics on the difference between observations and the first guess and analysis fields are routinely prepared to identify any problems with either the analysis system or individual data types. Lists of rejected data are also used to identify unreliable reporting from particular observing platforms.

There are three major operational analysis and forecast systems (viz. the global GASP, regional LAPS and the tropical TLAPS) in the NMOC Melbourne. A mesoscale version of LAPS, called MESO_LAPS, provides additional high resolution forecast products over a domain somewhat smaller than LAPS. The domains for each of these systems are shown in the figure below. The regional and tropical systems are dependent on the global system for their lateral boundary conditions, whereas MESO_LAPS is nested in LAPS. An additional system, called TCLAPS, is run whenever a tropical cyclone is present in the region. Manual

intervention is used for mean sea level pressure in both the global and regional systems. The resulting hemispheric "pseudo-observations" for mean sea level pressure are disseminated on the GTS. The tropical and Australian region limited area systems use bogus moisture data derived from GMS satellite imagery together with a tropical cyclone bogus scheme. An additional feature of the tropical system is its dynamical nudging. Output from the global system is also used in the cold start procedure for the Australian region and tropical systems. It is noted that MESO_LAPS system does not have its own separate analysis associated with it but uses initial (and boundary) conditions derived directly from LAPS.

The global, Australian region and mesoscale streams have associated sea-state systems. There are a large number of other parts to the basic scheme. These include systems for sea surface temperature analysis, environmental emergency response, generation of weather elements from model output (MOF), amendment and dissemination of aviation products, MCIDAS, archives, verification, display and dissemination of products. A schematic representation of the overall system is shown in the figure above.

At the present time, the centre produces major analyses at 00 and 12 UTC daily for the globe, Australian region and a tropical domain. Global forecasts out to 8 days, Australian region and tropical forecasts out to 48 hours, and mesoscale forecasts out to 36 hours, are produced off these major analyses. The ECMWF's Supervisor Monitor Scheduler is used to initiate and monitor the various tasks in the operational NWP suite. An approximate daily schedule for the main NWP systems is shown in the table (with the times during the daylight saving months, November to March, shown in brackets):

The acronym GASP is given to the Global ASimilation and Prognosis system. Post-processed products from this system are disseminated on the GTS in GRIB form, nationally through "DIFACS" and MCIDAS, and also via Radio-facsimile broadcasts (AXI and AXM).

Assimilated data: Mean sea level pressure (surface network, ships, drifting buoys), thickness (radiosondes, satellite retrievals), moisture (dew points, satellite precipitable water), wind (rawinsonde, aircraft, geostationary satellites, constant level balloons).
Assimilation cycle, including cut-off time: 6 hourly cycling. H+6 cut-off.
Method of analysis: Multivariate statistical interpolation + univariate O.I. for moisture.
Analysed variables: Geopotential, wind, moisture.
First guess: 6 hour forecast from previous cycle.
Coverage: Global.
Horizontal resolution: Triangular 239.
Vertical resolution: 29 sigma levels (0.991, 0.975, 0.950, 0.925, 0.900, 0.875, 0.850, 0.800, 0.750, 0.700, 0.633, 0.566, 0.500, 0.433, 0.366, 0.320, 0.290, 0.260, 0.230,
0.200, 0.170, 0.140, 0.110, 0.090, 0.070, 0.050, 0.030, 0.020, 0.010)
Initialization: Incremental non-linear normal mode.

Basic equations: Spectral primitive equations.
Independent variables: latitude,longitude,s,t
Dependent variables: log p*,T,q,vorticity,divergence.
Numerical technique:
horizontal: Spectral.
vertical: Finite difference.
time: Semi-implicit semi-Lagrangian.
Integration domain (in horizontal and vertical): Global, surface to 10hpa (approx.).
Horizontal and vertical resolution, time step: Triangular 239, 29 sigma levels, 600 sec.(approx.).
Orography, gravity wave drag: Both included.
Horizontal diffusion: Included.
Vertical diffusion: Included.
Planetary boundary layer: Included.
Treatment of sea surface, earth surface and soil: Included.
Radiation: Diurnal cycle, diagnostic clouds, interactive optical properties.
Convection (deep and shallow): Included.
Atmospheric moisture: Included.
Boundaries: Stand alone.

The following post-processed fields are available from GASP:

7.2.4     OPERATIONAL TECHNIQUES FOR APPLICATION OF NWP PRODUCTS:

The 10 m wind field from GASP is used to drive a global sea-state model.

7.3     SHORT-RANGE FORECASTING SYSTEM (0-72 HRS):

The Australian region Limited Area Prediction System (LAPS), the Tropical Limited Area Prediction System (TLAPS) and the Mesoscale Limited Area Prediction System (MESO_LAPS) provide forecasting guidance and products that are disseminated nationally through "DIFACS" and MCIDAS and also via Radio-facsimile broadcasts (AXI and AXM). TLAPS and TCLAPS, for specific tropical cyclone guidance, are run on behalf of RSMC Darwin. Again it is noted that, at present, MESO_LAPS does not have its own assimilation, or analysis, but uses an initial starting condition derived directly from LAPS. It is also noted that TCLAPS runs in 2 basic parts. The first preparatory part produces analyses and prognoses over a large domain, or what is called the Large Scale Environment (LSE), and the second part then generates analyses and forecasts at a higher resolution on a relocatable domain centred on the tropical cyclone (and nested within the LSE part).

Assimilated data: Surface synop, ship, drifting buoy, radiosonde, rawinsonde, GTS TOVS, locally processed TOVS, GTS and locally derived GMS cloud drift winds, aircraft single level winds, bogus MSLP ("pseudo-observations"), synthetic GMS moisture data, tropical cyclone bogus data.
Assimilation cycle, including cut-off time: 6 hourly cycling; cut-off: LAPS: H+2 hr, TLAPS: H+4 hr.
Method of analysis: Multivariate statistical interpolation + univariate statistical interpolation for moisture.
Analysed variables: Geopotential, wind, moisture.
First guess: 6 hour forecast from previous cycle.
Coverage: LAPS: 17.125⁰N-65.0⁰S, 65.0⁰E-184.625⁰E TLAPS: 44.25⁰N-45.0⁰S, 70.0⁰E-188.25⁰E
TCLAPS: 19.25⁰S-55.0⁰S, 75.0⁰E-159.75⁰E (LSE), 27.0⁰x27.0⁰ (Relocatable)
Horizontal resolution: LAPS and TLAPS: 0.375⁰
TCLAPS: 0.75⁰ (LSE) and 0.15⁰ (Relocatable)
Vertical resolution: LAPS, TLAPS, MESO_LAPS: 29 sigma levels (0.9988, 0.9974, 0.9943, 0.9875, 0.9750, 0.9625, 0.9500, 0.9250, 0.9000, 0.8750, 0.8500, 0.8000, 0.7500, 0.7000, 0.6000, 0.5000, 0.4500, 0.4000, 0.3500, 0.3000, 0.2750, 0.2500, 0.2250, 0.2000, 0.1750, 0.1500, 0.1000, 0.0700, 0.0500)
TCLAPS: 19 sigma levels (0.9910, 0.9750, 0.9500, 0.9000, 0.8500, 0.8000, 0.7500, 0.7000, 0.6000, 0.5000, 0.4000, 0.3500, 0.3000, 0.2500, 0.2000, 0.1500, 0.1000, 0.0700, 0.0500)
Initialization: LAPS and MESO_LAPS: Digital filtering technique.
TLAPS and TCLAPS: diabatic dynamical nudging scheme incorporating GMS IR imagery.

Basic equations: Grid primitive equations.
Independent variables: x,y,z,t
Dependent variables: P*,T,q,u,v.
Numerical technique:
horizontal: Finite difference.
vertical: Finite difference.
time: Semi-implicit.
Integration domain (in horizontal and vertical):
LAPS:17.125⁰N-65.0⁰S, 65.0⁰E-184.625⁰E, surface to 50 hpa (approx.)
TLAPS: 44.25⁰N-45.0⁰S, 70.0⁰E-188.25⁰E, surface to 50 hpa (approx.)
MESO_LAPS:4.875⁰S-55.0⁰S, 95.0⁰E-169.875⁰E, surface to 50 hpa (approx.)
TCLAPS: 19.25⁰S-55.0⁰S, 75.0⁰E-159.75⁰E (LSE), 27.0⁰x27.0⁰ (Relocatable); surface to 50 hpa (approx.).
Horizontal and vertical resolution, time step: LAPS and TLAPS: 0.375⁰ , MESO_LAPS: 0.125⁰, 29 sigma levels; TCLAPS 0.75⁰ (LSE) and 0.15⁰ (Relocatable), 19 sigma levels; LAPS and TLAPS: 40 sec., MESO_LAPS: 10 sec., TCLAPS: 40 sec (LSE) and 15 sec (Relocatable).
Orography, gravity wave drag: Included.
Horizontal diffusion: Included.
Vertical diffusion: Included.
Planetary boundary layer: Included.
Treatment of sea surface, earth surface and soil: Included.
Soil moisture analysis: Included in LAPS, TLAPS, TCLAPS and MESO_LAPS.
Radiation: Diurnal cycle, diagnostic clouds, interactive optical properties.
Convection (deep and shallow): Included.
Atmospheric moisture: Included.
Boundaries: LAPS, TLAPS and TCLAPS (LSE): Lateral boundaries from GASP.
MESO_LAPS: Lateral boundaries from LAPS.
TCLAPS (Relocatable): Lateral boundaries from LSE TCLAPS.

The following post-processed fields are available from the Limited Area Prediction Systems:

The 10 m wind field from LAPS and MESO_LAPS are used to drive sea-state models for the Australian region and smaller domains. The Model Output Forecast (MOF) system, which is driven by LAPS and GASP, is used to produce numerous weather elements including temperatures (minimum, maximum,dry bulb, dew point, wet bulb and ground minimum), wind (speed and direction), precipitation, cloud amount, evaporation, sunshine and visibility.

An analogue/regression tropical cyclone model ("TOPEND") is available for providing guidance to the Australian Tropical Cyclone Warning Centres. Specialized sea-state forecasts are provided for the North West Cape and Bass Strait gas and oil fields.

7.4.1 Sea Wave Models:

The following table summarises the characteristics of the seastate system in the NMOC:

7.4.2    Air Dispersion Model:

An air dispersion, or transport, system is available for running on demand, in an ad-hoc mode, and can produce forecast trajectories, concentrations (or exposures) and depositions for nuclear accident, volcanic ash, smoke and other episodes. Currently, the operational Environmental Emergency Response system consists of the Hybrid Single_Particle Lagrangian Integrated Trajectories (HY-SPLIT Version 4.0) system, developed at the NOAA Air Resources Laboratory, with meteorological input from the operational NWP systems in NMOC (viz., GASP, LAPS, TLAPS, and MESO_LAPS). Analysed backward trajectories are also available from GASP, LAPS and TLAPS in operations.

7.4.3    Solar Ultraviolet (UV) Radiation Forecast System:

Forecasts out to 36 hour forecasts of an UV index (defined as the product of the UV irradiance and a human skin response function) are produced operationally. The system analyses ozone concentrations, available from the global 120 km GTS TOVS data, using 2-dimensional univariate statistical interpolation. Forecasts of the ozone distribution are then computed using isentropic fields derived from GASP output. From vertical profiles of temperature and ozone, the UV index is calculated. Currently, forecasts of the UV index are produced, once per day, after completion of the 1200 UTC run of GASP.

7.4.4    Storm Surge System:

A system capable of forecasting storm surges caused by tropical cyclones is currently driven by wind and pressure field distributions defined according to the central pressure and maximum wind radius bogussed for tropical cyclones by Darwin RFC.

7.5     EXTENDED-RANGE FORECASTS (10 - 30 DAYS): Not applicable yet.

A three-month rainfall seasonal climate outlook is prepared. Each month, a risk-assessment for three-month total rainfall across Australia is issued mid-month for the three-month period starting the following month. Probabilities are calculated for the three-month total rainfall being in the lowest one-third of historical falls (tercile 1), the middle one-third (tercile 2), and the upper one-third (tercile 3). The technique used is discriminant analysis, with the inputs being derived from recent Sea Surface Temperature (SST) patterns. Subsidiary techniques involved in the forecast model include principal component analysis of SSTs for the Pacific Ocean, Indian Ocean and Southern Ocean, and principal component analysis of rainfall patterns across Australia. SST EOF (Empirical Orthogonal Function) loadings, at one and three months lag, for the Pacific Ocean ENSO pattern and the Indian Ocean pattern are the current predictor inputs. The tercile probabilities, computed across Australia on a 1⁰x1⁰ grid are published in the form of contoured maps, tabulated averages for the 107 Australian rainfall districts, and tabulated interpolations for cities and towns around Australia.

Additional guidance at the rainfall district level is presented in the form of stratified rainfall climatologies based on recent values of the SOI (Southern Oscillation Index). Rainfall outcomes for eastern Australia, obtained from SOI analogues, are also described.

A similar scheme has been developed for seasonal temperature forecasts within the Bureau's research centre (BMRC) and this will eventually become operational. Additional forecasts for NINO3 are provided based on an intermediate coupled atmosphere-ocean with sub-surface ocean temperature assimilation. A coupled atmosphere-ocean GCM (General Circulation Model) combined with a sub-surface ocean temperature data assimilation system has been developed and is currently being assessed as a more comprehensive seasonal forecasting tool.

An annual summary of verification statistics for 1999, for the Australian regional and global schemes, is given in the following tables.

The following abbreviations have been used in the tables:

Note:

REGN refers to LAPS (Limited Area Prediction System).

The LAPS and GASP results are with respect to their own analyses.

24 HR VERIFICATION STATISTICS - AUSTRALIAN REGION 1999

BASE TIME: 0000UTC - VALID TIME: 0000UTC

BASE TIME: 1200UTC - VALID TIME: 0000UTC

BASE TIME: 0000UTC - VALID TIME: 0000UTC

9. PLANS FOR THE FUTURE:

Plans for future operational systems in the NMOC Melbourne include:

• an ongoing improvement and generalization of data processing;

• the introduction of a higher horizontal resolution, viz. 0.05⁰, version of MESO_LAPS for the Syney and Melbourne areas;

• the implementation of an associated air quality prediction system;

• an upgrade to a statistical interpolation version of the rainfall analysis system;

• the introduction of a rainfall verification scheme;

• the introduction of a mesoscale assimilation system;

• an ongoing improvement of graphical display and editing systems;

• the incorporation of local direct readout TOVS data, locally derived cloud drift winds and ERS scatterometer winds into GASP;

• the implementaion of the 1D-VAR system into GASP;

• a move to drive the storm surge model using meteorological input from TLAPS;

• a move to produce the 96-hour global seastate forecasts at an improved resolution of 1.0⁰;

• an ongoing improvement to the operational Environmental Emergency Response System including the use of a unified wind input system and an extension to include a number of chemical transformations;

• a move to broad-scale guidance on the 1-2 week time-scale;

• the introduction of a fully coupled ocean-atmospheric model for seasonal analysis and prediction;

• an upgrade to a dual-node NEC SX-5.

Bourke W., Hart T., Steinle P., Seaman R., Embery G., Naughton M. and Rikus L. : "Evolution of the Bureau of Meteorology's Global Data Assimilation and Prediction System. Part 2: Resolution Enhancements and Case Studies." Australian Meteorological Magazine, 44, 19-40, 1995.

Davidson N.E. and Puri K. : "Tropical prediction using dynamical nudging, satellite defined convective heat source and a cyclone bogus." Monthly Weather Review, 120, 2501-2522, 1992.

Davidson N.E. and Webber H.C.: "The BMRC high resolution tropical cyclone prediction system TC-LAPS." To appear in Monthly Weather Review, 2000.

Draxler R.R. and Hess G.D. : "An overview of the HYSPLIT_4 modelling system for trajectories, dispersion and deposition." Australian Meteorological Magazine, 47, 295-308, 1998.

Le Marshall J., Pescod N., Seaman R., Mills G. and Stewart P. : "An Operational System for Generating Cloud Drift Winds in the Australian Region and Their Imapct on Numerical Weather Prediction." Weather and Forecasting, 9, 361-370, 1994.

Le Marshall J.F., Riley P.A., Rouse B.J., Mills G.A., Wu Z.-J., Stewart P.K. and Smith W.L.: "Real-time assimilation and synoptic application of local TOVS raw radiance observations." Australian Meteorological Magazine, 43, 153-166, 1994.

Puri K., Dietachmayer G.S., Mills G.A., Davidson N.E., Bowen R.A. and Logan L.W. : "The new BMRC Limited Area Prediction System, LAPS." Australian Meteorological Magazine, 47, 203-223, 1998.

Seaman R., Bourke W., Steinle P., Hart T., Embery G., Naughton M. and Rikus L. : "Evolution of the Bureau of Meteorology's Global Data Assimilation and Prediction System. Part 1: Analysis and Initialisation." Australian Meteorological Magazine, 44, 1-18, 1995.

Stewart P.K. : "RSMC Melbourne - Pocedures, Standards, Results of Exercices and Ongoing Products." WMO Expert Team on Environmental Emergency Response. Beijing, China, 20-24 September 1999.