Model Development Group

Group Leader: K. Puri

Objectives: To conduct research on atmospheric modelling to support weather and climate research and operations.

Research activities

Unified BMRC Atmospheric Model (BAM)

Significant progress has been made in upgrading the BMRC Atmospheric Model (BAM) with enhanced physics and computational efficiency. New developments include completion of the global model’s interface to the European Centre for Medium-Range Weather Forecasts (ECMWF) land surface/boundary-layer/vertical diffusion scheme; the bulk explicit microphysics scheme (BEM); implementation of the Lott-Miller gravity wave drag formulation; capability for interfacing to the single-column model; performance optimisation in a number of routines; a complete overhaul of the code to generate surface fields and a new treatment of surface roughness over land. A number of Atmospheric Model Intercomparison Project (AMIP)-type runs have been performed to evaluate some of the major changes, including the testing of the ECMWF package, the Lott-Miller gravity wave drag and the Sun-Edwards-Slingo (SES) radiation scheme. Additional runs have been carried out to test a number of options for the physical parametrisations and varying horizontal and vertical resolutions.

A significant amount of effort was put into developing Message-Passing Interface (MPI) versions of both the Global Assimilation and Prediction System (GASP) and the Limited Area Prediction System (LAPS). Additional effort was made to maximise the vectorisation and parallelisation in the code which resulted in a high level of performance in both models. Implementation of MPI in the LAPS Eulerian code has been completed and work is in progress to implement in the semi-Lagrangian version of the global model.

BAM4.0, which includes all the above upgrades, was officially released in December 2003 after significant testing in the climate and numerical weather prediction (NWP) modes in the global model and in LAPS.

Parametrisation of physical processes

There are a number of ongoing activities in the areas of convection, diagnostic and prognostic clouds, radiation and land surface models. A Bulk Explicit Microphysics (BEM) scheme has been developed and implemented in BAM which includes a warm-rain scheme, prediction of cloud fraction, ice-phase microphysics and fallout of hydrometeors. A number of BEM-derived diagnostics (lightning prediction etc.) have also been developed. The scheme has been extensively tested through parallel runs with LAPS and the results show a clear positive impact on precipitation scores. The scheme was been implemented operationally in all versions of LAPS after extensive trials by the National Meteorological and Oceanographic Centre (NMOC). The hook up to the global model has been completed and detailed forecast-only tests in GASP have been completed.

A number of refinements have been made to the SES radiation scheme, including the replacement of ozone cross-section absorption data with new measurements and a new solar source function at the top of the atmosphere. Detailed testing of the SES radiation in conjunction with the ECMWF land surface/boundary-layer/vertical diffusion has been carried out in the climate version of the model. These experiments show a clear positive impact of the SES scheme over the Fels-Schwarzkopf radiation. Detailed testing of the SES scheme in the NWP mode in both GASP and LAPS will commence when the new supercomputer becomes available.

Work is ongoing on a number of convection-related activities. These include implementation of different closure, entrainment and microphysics options, strengthening of the coupling of convection with the rest of the physics, and investigation of the effects of convection on IR brightness temperatures. There has been a major effort to restructure the cloud, convection and radiation parametrisations to make them more modular, and to implement outfield modules to facilitate diagnostic field output from BAM.

A significant amount of effort has been made to enable the ECMWF land-surface model/boundary-layer/vertical diffusion scheme to be used in the global model. This work has been completed and testing of the scheme has been carried out in both the climate mode and GASP. These tests have confirmed the robustness of the scheme and the results indicate that it can now be used routinely. The Group’s experimentation will now increasingly use this scheme instead of the currently used bucket hydrology and the Louis planetary layer scheme. In addition to the above testing a significant upgrade has been made to the generation of input data for the scheme. This data, which includes bare albedo, vegetation fraction and vegetation roughness, has in the past been derived from the annual mean climatology of Wilson and Henderson-Sellers. The upgraded version now uses the French Ecoclimap dataset which is a 12-month climatology and therefore allows the use of seasonally varying soil and vegetation parameters. In addition it includes a number of soil and vegetation variables that allow the use of spatially variable soil and vegetation parameters. All necessary work to enable the use of this dataset in BAM has been completed and tested, including climate integrations.

The Lott-Miller gravity wave parametrisation has been successfully implemented in BAM and detailed comparisons with the current Palmer scheme have commenced. Two-year runs with both schemes have been completed. It is now being used in all climate version runs. The Lott-Miller scheme has also been tested in NWP mode in GASP.

An important initiative in the parametrisation area is the development of a single-column model (SCM). Implementation of the BAM physics within the National Center for Atmospheric Research (NCAR) SCM framework has been completed and successfully integrated with BAM4.0 physics.

Generation of model diagnostics and validation

Version 2 of the Model And Climatological data Comparison System (MACCS) has been released for general use. A number of additional fields have been implemented and it is now possible to derive seasonal and annual means with output as a HTML report that can be accessed by web browsers. Software to generate Taylor diagrams has been developed and tested.

A feature which has proved to be extremely useful to understand/debug different processes at a number of modelling centres is the ability to output and display variables during model integration. This facility (outfield modules) has now been successfully developed and implemented in BAM4.0. Software to output time series at selected points to interface with the ARM Program and the Global Energy and Water Cycle Experiment GEWEX Coordinated Enhanced Observing Period (CEOP) has been developed within BAM. A preliminary Model Output Location Time Series (MOLTS) dataset has been prepared for CEOP.

A significant amount of work is ongoing on cloud and radiation validation. Of particular note is the extensive development of synthetic satellite imagery from model output. Pseudo-operational synthetic satellite imagery for most operational NWP models is made available every 12 hours on the internal web, imagery has been extended to sub-regions of mesoLAPS, and work has commenced on developing downscaling routines for subgrid scale investigation.

In addition to the above, comprehensive validation packages exist for NWP applications. These include generation of standard measures such as skill scores, rms errors, anomaly correlations, biases; validation of weather elements such as rainfall, screen level temperatures (maximum/minimum temperatures), 10-metre winds; and tropical cyclone (TC) track/intensity errors. Software for validation of weather elements has been handed over to NMOC for operational implementation.

Comprehensive diagnostics for the GASP and LAPS ensemble prediction systems have been developed. Of particular note is the implementation of probabilistic precipitation verification developed by Ebert and probabilistic verification of maximum and minimum 2-metre temperatures and 10-metre winds developed by Yang and Ebert. This activity is expected to grow with the continuing development and application of these two new systems.

Operational developments

A significant amount of parallel running has been performed to validate new developments, such as the testing of the Bulk Explicit Microphysics (BEM) scheme in LAPS. The system was implemented operationally in LAPS375 in March 2003, followed later by mesoLAPS, TCLAPS and the 0.05^o version of LAPS used in the Australian Air Quality Forecasting System (AAQFS).

TCLAPS, the high-resolution version of LAPS developed for TC forecasting, continued to perform well and to be well received by forecasters. Verification of the 2003 season Australian region cyclones shows track errors of 149km and 253km at 24h and 48h respectively - these errors compare very favourably with errors from other models. Operational upgrades to TCLAPS include improvements in the specification of the initial vortex, increasing the number of vertical levels from 19 to 29 (which brings it in line with other configurations of LAPS) and implementation of BEM.

The capability to transport an arbitrary number of passive scalars has been added to LAPS. Apart from providing a good test-bed to test conservation properties in LAPS, this addition could be useful for other transport related issues such as smoke and dust transport.

The dynamics core (DC) of LAPS is now moderately old and could benefit from a re-write which would include a nonhydrostatic capability. To meet this requirement a minimalist strategy will be used in the shorter term. This phase will maintain as much of the basic DC as possible, construct the nonhydrostatic equations using the Laprise ‘hydrostatic pressure’ coordinate, and introduce an implicit treatment in the vertical to prevent instability associated with vertically propagating sound waves. Phase 2 will involve a complete redesign of the dynamical core.

Data assimilation using new data sources such as scatterometer winds, surface winds, satellite-derived heating rates, and US Naval Research Laboratory (NRL) rain rates has been conducted within the framework of LAPS. Preliminary results suggest that assimilation of these data could have significant impact on the prediction of mesoscale features including intensification of tropical lows. This system has now been developed into a prototype of Severe-Weather LAPS (SWLAPS) and detailed testing of its performance has commenced. SWLAPS is designed specifically to conduct additional runs of LAPS to provide additional guidance during severe weather situations. The initial configuration of the system consists of a horizontal resolution of 0.1^o with 29 vertical levels. This development is being carried in collaboration of the Data Assimilation Group.

An ensemble prediction system based on LAPS has been developed which uses random perturbations of observations during data assimilation, stochastic physics and different options for convective parametrisation to generate the ensemble members. An important feature of the system is applications to tropical cyclone track prediction, where in the presence of a cyclone the TC bogus data is perturbed by varying the position, size and motion of the cyclone.

The Australian Air Quality Forecasting System (AAQFS) is a collaborative project between the Bureau of Meteorology, CSIRO Atmospheric Research, CSIRO Energy Technology, Environment Protection Authority of Victoria, and the New South Wales Environment Protection Authority. As the project expands, EPAs from other states will also be involved. Under the project, a 0.05^o version of LAPS (LAPS05) has been run twice daily in the Melbourne domain since January 1998 and in the Sydney domain since June 1998. The hourly output is used to drive the CSIRO chemical transport model (CTM). A comprehensive evaluation of the system during the summer oxidant season 2001-2002 in both Sydney and Melbourne has been completed. The emphasis of this study was to explore first steps towards determining the limits of predictability by investigating the effect of the size of the forecast domain, the time of initiation and length of the forecast period and the complexity of the chemistry employed.

Long-range transport of bushfire smoke from King Island to Melbourne was the basis of another study. The results from the models for AAQFS and HYSPLIT were compared and showed the ability of both systems to perform well.

A wind-blown dust module was implemented for daily operation in the chemical transport model (CTM) for the whole of Australia and both the Victoria and New South Wales domains. An automated satellite-initiated plume-envelope model of bushfire smoke was implemented for daily operation for Victoria and New South Wales. A sensitivity study is underway to determine the contribution to PM10 from various sources: paved-road dust, domestic wood combustion and wind-blown dust. The emissions inventory has been re-written for efficiency and robustness. Scaling has been introduced for motor vehicle emissions by process (exhaust, evaporative, road dust) and fuel type (petrol, diesel, LPG). The biogenic module has been updated.

Testing of the ECMWF land-surface scheme together with nudging of soil moisture has been performed by carrying out data assimilation (using GenSI) and 5-day forecasts for the October-November 2003 period. Key verification results are a major positive impact in the tropics, generally positive impact in the southern hemisphere and Australian region, and neutral or just slightly worse in the northern hemisphere. Forecast only runs of GASP with BEM have also been conducted which indicate satisfactory performance of BEM in GASP.

The GASP Ensemble Prediction System has been running in NMOC operational trial since July 2001 and a comprehensive list of diagnostics and verifications from the system has been developed. Additional verifications such as the probablistic precipitation verifications have been implemented. The system is now a part of NMOC’s routine medium-range forecast guidance suite.

Techniques development

The Environmental Emergency Response (EER) transport and dispersion modelling efforts are based on the HYSPLIT model. A model upgrade has been developed (Version 4.7) and implemented recently. Particle and puff mixing routines have been recoded to use the turbulent velocity variance rather than the diffusivity to compute the dispersion more directly. New equations have been added to directly compute the turbulent velocity variances from surface/boundary layer stability. The graphics programs have been modified to improve the display of short trajectories and plumes, and trajectory points can be plotted at intervals of less than one hour. These upgrades give greater flexibility in the use of the model for forecasting transport and dispersion of volcanic ash, bushfire smoke, foot-and-mouth disease virus, radioactive plumes, dust storms, and other applications. Applications over the past year have included:

· Collaboration with Department of Agriculture, Fisheries and Forestry in Canberra on development of a foot-and-mouth protocol for producing warnings for a Foot-and-Mouth Disease Virus Warning System

· Participation in the International Environmental Emergency Response (EER) Monthly Tests with Regional Specialized Meteorological Centres (RSMCs) in Montreal and Washington, DC

· Participation in the International Test of the Comprehensive Test Ban Treaty Organisation (requiring retro-dispersion plumes to locate possible source regions for clandestine nuclear explosions)

· Simulations of volcanic ash and bushfire smoke transport and dispersion inter-comparison and verification of AAQFS and HYSPLIT for long-range bushfire smoke transport from King Island

· Simulations of dust storms and inter-comparison and verification of AAQFS and HYSPLIT for the major dust storm event of 23 October 2002

A second Techniques Development project is the Ultraviolet Radiation and Ozone analysis and forecasts system. A number of improvements were implemented this year. A new scheme using blended (TOVS + TOMS) ozone was implemented operationally in July 2003 following extensive trials. The system is being validated through daily comparisons of model clear sky UV Index, clear sky UV dose and total ozone amounts with TOMS observations. Comparison with Australian Radiation Protection and Nuclear Safety Agency (ARPANSA) UV measurements has been performed for the period 1997-2000. New products from the system include daily UV and ozone for different locations made available via the internal web page, daily total erythemal dose and daily clear sky UV Index distribution for Australian cities following the international standards recommended by the World Health Organization (WHO). This product, which involved close collaboration with the Cancer Councils of Victoria and New South Wales, ARPANSA, and the Weather and Ocean Services Policy Branch, is generated daily and is made available on the Bureau’s external web page. Global and regional model analyses and forecasts for clear and cloudy sky UV Index have been made available to a number of registered users (Israel, Singapore, Cancer Council, ARPANSA, University of Tasmania, University of Colima Mexico).