Chapter 3 Climate Applications and Services

Climate Services > Climate Prediction

The Bureau of Meteorology’s NCC has, since 1989, routinely provided guidance as to expected seasonal conditions over Australia. The publicly distributed form of this advice centres on eight rainfall and eight temperature Seasonal Outlook Statements (national and one for each State and the Northern Territory) disseminated monthly and covering the subsequent three month period. These are complimented by a large number of web pages and brochures, which provide background to Australia’s climate. This public service is supported by a more detailed monthly publication, the Seasonal Climate Outlook (SCO), which is available for the cost of provision through subscription. The SCO gives probabilities of reaching specified rainfall amounts over standard rainfall districts and for selected towns over a three month period together with descriptions of recent climate anomalies over Australia, and long term projections of the El Niño–Southern Oscillation phenomenon. The booklet is available either by hardcopy or electronically through the Bureau of Meteorology’s SILO web site (www.bom.gov.au/silo).

Advances in understanding of the global climate system and the coherence of seasonal to interannual variability patterns have provided the basis for prediction on seasonal time-scales. Monitoring of the evolution of El Niño/La Nina events, and the Southern Oscillation Index (SOI) in conjunction with sea-surface temperature analysis generally provides the foundation for current statistical predictions of climate in the Australian region.

Complementing releases of climate outlooks, the Bureau of Meteorology Research Centre (BMRC) also issues statements targeted to agricultural and water agencies in Western Australia regarding the prospects for rainfall and temperature in the southwest of that state. These statements are based on a regionalised version of the NCC’s statistical prediction model. These statements are designed to assist in water management decisions over the summer months.

Recent years have seen significant effort in developing climate prediction on both the intraseasonal (less than 3 months) and longer time scales (beyond 3 months). On the longer time scales, statistical seasonal outlooks are now augmented with forecasts from dynamical models which are operationally integrated to provide forecasts for as long as a year ahead. Late 2002 saw the Bureau of Meteorology introduce an operational dynamical ensemble prediction system (Predictive Ocean Atmosphere Model for Australia – POAMA) using model components developed collaboratively by CSIRO Marine Research and the Bureau of Meteorology Research Centre. While this system remains a matter of ongoing scientific research and development, forecasts from this model have been very successful to date, predicting both the continuation of the moderate El Niño event through 2002, and its eventual demise in autumn 2003 (Figure 3.2).

Figure 3.2 Observed (solid) and predicted (dashed) sea surface temperatures in the central tropical Pacific (NINO 3.4) during the 2002-03 El Niño event. The prediction "plumes", originating from spring 2002, are from the Predictive Ocean Atmosphere Model for Australia (POAMA) and show a successful prediction of the demise of the El Niño event.

Along with ocean forecasts derived from POAMA, the NCC provides a summary of (and links to) long-range forecasts from a total of 12 different long-range forecast systems. These climate predictions are drawn from such groups as CSIRO Atmospheric Research and the European Centre for Medium Range Weather Forecasts. This broader perspective has proven useful in illustrating the uncertainties involved in long-range forecasts, and effectively provides a subjective multi-model ensemble. This summary is made available to the public through the Bureau of Meteorology’s web site.

CSIRO have also developed a seasonal forecast model, funded by the Climate Variability in Agriculture Program (CVAP) LWRRDC), and which has been running operationally since 2001. It is a global coupled atmospheric-oceanic model, based on the CSIRO Mark 2 model, and is one of a very small group of global coupled dynamical models in use or under development worldwide.

The Coupled OASIS CSIRO9-Mark 2 ACOM2 (COCA) model is run operationally at least once per month and issues realtime predictions of the likelihood of El Niño or La Niña developments and estimates of rainfall probabilities for Australia up to 6 months ahead. The COCA El Niño predictions are publicly available via the Internet (http://www.dar.csiro.au/climate/coca.html), are supplied to the International Research Institute for inclusion in the summary of international prediction groups ( http://iri.columbia.edu/climate/ENSO/currentinfo/SST_table.html) and are reported in the NCC Seasonal Climate Outlook.

The development of a new seasonal prediction model based on the much improved CSIRO Mark 3 model is now nearing completion. The next phase will see the development of an initialisation technique applicable to multi-member ensemble predictions, a necessity resulting from the inescapable uncertainty arising from atmospheric chaos. The third phase will use output fields from the Mark 3 global model to drive regional downscaling models, which will greatly enhance the skill of the climatic predictions, especially rainfall.

CSIRO has also developed a statisticallybased agricultural prediction system based on ocean temperatures. The ‘Oceans to Farms’ system uses historical links between ocean temperature patterns and variables linked to agricultural output to provide tailored, vertically-integrated forecasts for specific regions and industries. For example, ocean temperatures have been used to forecast growing days and this forecast, in turn, is used in an agricultural model to optimise farm gate output nine months later. In north Queensland this system outperforms even perfect knowledge of rainfall, because it takes into account other factors relevant to plant growth. The CSIRO forecast system is unique because it can be tailored to particular industries and regions.

In order to make best use of the seasonal outlook information available, users must consider how well the outlook performs for their particular area. Scientific measures of skill, whilst providing detailed information to the researcher, are generally not userfriendly. The NCC, in collaboration with CVAP, has been examining how to best communicate this information to the typical Seasonal Outlook user. The result of this collaboration is a web page, containing Australia-wide maps of "Percent Consistent", which gives the percentage of years in which the category (above or below normal) favored by the probabilities in the outlook was subsequently observed. Users can make a detailed assessment of the reliability of the seasonal outlook for their region. Work is underway to extend this web site to provide verification information for recently produced climate forecasts. The current statistical prediction scheme has good skill over much of the continent (Figure 3.3) and very useful rainfall and temperature (Figure 3.4) forecasts were issued during the 2002 drought.

Figure 3.3 A measure of the consistency of forecasts for above/below median seasonal maximum temperature forecasts issued by the Bureau of Meteorology since commencing the operational service based on sea surface temperatures (autumn 2000 to summer 2002-03).

Figure 3.4 The Bureau of Meteorology’s Seasonal Climate Outlook system performed well during the 2002 drought. The temperature outlook for October to December issued in September (top) indicated a strong likelihood of above average maximum temperatures over much of the continent. The actual temperatures observed for this period (bottom) were generally consistent with this forecast.

A component of these verification activities will see the Bureau of Meteorology encourage and support both Australian and international producers of seasonal to interannual predictions to document climate forecast skill, following the model of the WMO Commission for Basic Systems (CBS) Standardised Verification System (SVS) for Long-Range Forecasts (LRF). The Bureau of Meteorology was also involved in meetings to discuss WMO’s plans for improved infrastructure for seasonal-to-interannual prediction over the globe. It is expected that the future will see a number of Regional Climate Centres (RCCs) taking data and products from major "prediction-producers", termed Global Producing Centres, and providing tailored products and services to National Meteorological and Hydrological Services (NMHSs). RCCs, which may be virtual entities, will be expected to provide various other climate services.

The Bureau of Meteorology Research Centre has been investigating techniques to monitor and predict on intraseasonal timescales. An experimental monitoring/forecast system for equatorial regions has been established and is being tested. Preliminary results from this scheme indicate that precipitation variations and the onset of active and break monsoon conditions can be skillfully forecast out to about 15-20 days. The Bureau of Meteorology’s Northern Territory Regional Office introduced a Weekly Tropical Cyclone Formation Forecast based on the phase of the intraseasonal oscillation and giving probabilities of tropical cyclone formation in the Australian region for the subsequent five weeks.

Seasonal rainfall forecast guidance is produced by the Queensland Centre for Climate Applications (QCCA), which is a joint effort between the Queensland Department of Primary Industries (QDPI) and the Queensland Department of Natural Resources and Mines (QDNR&M). These are based on the historical relationship between the "phase" of the Southern Oscillation during the most recent two months, and subsequent rainfall anomalies, and made available in real-time on the QCCA web site and in the media.

The West Australian Department of Agriculture has been involved in the development of an El Niño prediction system based on mid-latitude pressure indices, which have the potential to offer a longer lead-time than systems based on indices developed from tropical data. This system is linked to a crop yield forecasting model.

A major activity of the Climate Impacts and Natural Resource Systems (CINRS) group in the QDNR&M has been the operational Drought and Degradation Alert System, which is supported by a range of collaborative activities, including:

• development of better climate forecasting systems customised for management decisions in Queensland’s grazing lands and Australia’s rangelands;
• simulation case studies of grazing enterprises evaluating the use of climate forecasting for grazing land management in terms of animal production and resource degradation risk;
• evaluation of the application of General Circulation Models in climate forecasting, land management feedbacks and climate change;
• analysis of potential climate change impacts on Queensland’s grazing lands;
• integration of climate driven models, remote-sensing and extensive groundtruthing to provide improved reporting of resource status in terms of land cover and carbon stocks; and
• delivery of products through the Long Paddock web site (www.longpaddock.qld.gov.au).