Date Time Title Speaker Affiliation   2   February (Mon) 10am Application of GPS radio occultation data to weather analysis and prediction and the COSMIC mission Richard Anthes, Bill Kuo, and Christian Rocken UCAR, USA   3   February (Tue) 10am The use and impact of automated aircraft observations in data assimilation: cruise and ascent/descent phases Professor Joel Tenenbaum State University of New York, USA   5   February (Thu) 10am Recent developments in Doppler radar data assimilation Qin Xu NOAA, USA 10 February (Tue) 10am Gravity wave and convection research at the University of Adelaide Professor Robert Vincent University of Adelaide 24 February (Tue) 10am Circulation regimes and SST forcing: Results from large GCM ensembles David Straus Centre for Ocean-Land-Atmosphere Studies 25 February 10am Recent developments in atmospheric modelling as reported at the WGNE session. Discussions on future scientific directions and challenges in WCRP Kamal Puri BMRC   4   March (Thu) 10am Multiple-sensor severe weather application development at NSSL Greg Stumpf NOAA/NSSL 10 March 10am The major Tropical Western Pacific cloud regimes and their cloud and radiative characteristics Christian Jakob BMRC 17 March 10am Intraseasonal variability of the Australian-Indonesian monsoon region Matthew Wheeler and John McBride BMRC 19 March 10am Long term memory of the climate system: Observations, simulations and concepts Professor Klaus Fraedrich University of Hamburg 31 March 10am Stochastic variability associated with ENSO Harry Hendon BMRC   7 April   10am Revisiting the meteorology of Ash Wednesday Graham Mills BMRC     Easter Break     21 April 10am Towards assimilation of local ATOVS data in LAPS Chris Tingwell BMRC 28 April 10am Tropical Warm Pool International Cloud Experiment: An extravaganza in Darwin in 2006 Peter May BMRC   5 May   10am The Pyramid of CEOP Lawrie Rikus BMRC 10 May 10am Tropical cyclone research at the CRC (Cyclone Research Cell) of Meteo-France in La Reunion island Professor Miloud Bessafi Universite de La Reunion 12 May 4pm Context and climate change: Lessons from Barrow, Alaska Professor Amanda Lynch Geography and Environmental Science, Monash Uni 21 May 10am Polar vortices - observations and simulations - a GRIPS report Greg Roff BMRC   26 May   10am Statistical prediction of the intraseasonal modulation of tropical cyclone activity Anne Leroy BMRC/METEO-France student   2 June   10am Fog forecasting at Perth Airport using the Meso_LAPS model - guidance and verification Xinmei Huang BMRC   7   June 11:15am Weather-band circulation and upwelling off South Australia: a numerical study of the summer of 1999 John Middleton UNSW   9   June 10am UV and Ozone analysis and forecast system upgrade (2004) Lilia Deschamps BMRC 16 June 10am Ensemble single column modelling at the tropical Western Pacific ARM sites Tim Hume BMRC 24 June 10am Development of GenSI Peter Steinle BMRC 30 June 10am Bringing up 'El Niño': Rainfall outlooks in print media during 2002 Clare Mullen NCC 25 August 2pm Recent improvements to the NOAA AVHRR SST product at the Australian Bureau of Meteorology Anthony Rea OEB 30 August (Mon) 2pm Could El Nino be triggered from the South China Sea? Walter E. Janach Lucerne University of Applied Sciences, Switzerland   8   September 2pm Modelling and forecasting rainfall in space and time Alan Seed BMRC 22 September 2pm Advances in the provision of warnings for volcanic ash for aviation in the Australian region Rod Potts BMRC 29 September 2pm Supercomputing upgrade at the Australian Bureau of Meteorology Ilia Bermous BMRC 30 September (Thu) 10:30am The R&D Landscape: Implications for government research agencies John Walker Intellectual Property Management 6 October 2pm Dynamical patterns and synoptic tracks associated with different types of cold air surges Over South America Alex Pezza Earth Sciences, University of Melbourne 13 October 2pm Hourly operational consensus forecasts Chermelle Engel BMRC 20 October 2pm Climate change and Australia's fauna and flora Lynda Chambers BMRC 27 October 2pm Ocean model and assimilation developments Gary Brassington BMRC 28 October (Thu) 10am Features of C-CAM dynamics and plans for G-CAM John McGregor CSIRO/CAR 3 November 10am The Joint Centre for Satellite Data Assimilation: Goals and Achievements John Le Marshall JCSDA 4 November (Thu) 10am Issues in the design and implementation of an Ensemble Kalman Filter Jeff Kepert BMRC 8 November (Mon) 2pm Vertical level placement in GCMs - does it matter? Greg Roff BMRC 12 November (Fri) 10am Land data assimilation Paul Houser NASA-GSFC 22 November (Mon) 2pm Some results from the Radiation Olympics IRS2004: Water vapour on Tibetan Plateau, MODIS products and aerosol modelling Zhian Sun BMRC 23 November (Tue) 2pm Ground-based measurements of the radiative properties of the atmosphere - an ARM perspective Tom Ackerman Pacific Northwest National Laboratory 29 November (Mon) 3pm POAMA Seasonal Prediction System: Past, Present and Future Oscar Alves BMRC Week 6-10 December   BMRC Workshop     15 December 11am Improving the simulation of tropical Pacific climate: The impacts of biology and a more accurate surface stress calculation Noel Keenlyside Leibniz-Institut fuer Meereswissenschaften 22 December 2pm Using a knowledge based forecasting system to establish the limits of predictability Harvey Stern VRO/CCS

The temporary venue is the BMTC main lecture room (Floor 9, north side).

Until further notice, seminars are at 2 pm on Wednesdays with duration of 30 to 50 minutes + questions. Dates and times other than the usual Wednesday afternoon are shown in bold print.

Emphasis is on work in progress. Partly because of this, the schedule is susceptible to change.
To receive this information via email, simply send an email to majordomo@bom.gov.au containing the single line in the body of the message: subscribe bmrc_seminars

Videotapes of several seminars that have been given during the year are available for loan from the National Meteorological Library. These are indicated by a camera icon next to the seminar date. In addition, a list of seminars held in the library can be found on the catalogue by entering Series: BMRC, Format: Video. If you would like to have a talk videotaped please contact the seminar coordinator.

If you would like to know more details of coordinating seminars (if, for example, you are hosting a visitor who will be giving a seminar and the regular seminar coordinator is not available), have a look at the document, "Instructions for BMRC Seminar Coordinator"

For further details contact the seminar coordinator, Brett Harris, on 03 9669 4388, b.harris@bom.gov.au
or Lilia Deschamps, on 03 9669 4460, l.deschamps@bom.gov.au

ABSTRACTS

Radio occultation measures phase and amplitude of the microwave signals emitted from GPS. These signals are inverted to obtain profiles of signal bending, atmospheric refractivity, pressure, temperature and water vapor. The main purpose of the upcoming COSMIC mission is to demonstrate the value of these radio occultation products for weather forecasting and climate monitoring. This presentation will provide an overview of the COSMIC program, the status of data inversion, and issues related to assimilating this new data type into weather models. It will also show some recent results applicable to climate studies.

For more information on COSMIC please see http://www.cosmic.ucar.edu.

The previous efforts, however, were focused mainly on retrievals with zero background information. When these methods are upgraded for Doppler radar data assimilation, it is necessary to utilize the background information provided by model predictions. Ideally, the background information should be introduced in consistency with the general formulation derived for the conditional maximum likelihood estimate. How should this be done? Specifically, how should the general formulation be simplified (with certain assumptions) into a form suitable for assimilating Doppler observations? These and related technical problems (including data quality controls) will be addressed. Examples and results of numerical experiments will be presented to show how the previous developed simple adjoint method and least-squares method are upgraded for Doppler radar data assimilation.

Using a partitioning algorithm applied to each winter's ensemble separately, we find clusters in the 200 hPa height field which are significant (vis-a-vis a suitable Gaussian background), reproducible (in half-length data sets), and consistent (with clusters obtained from the 200 hPa u wind) for all winters except the strong El Nino events of 1982/83, 1986/87 and 1997/98.

One cluster found consistently in many winters, consisting of a strong ridge over the Alaskan region and a trough over central North America, is quite similar to the Alaska pattern identified from observations (Renwick-Wallace, 1996) as being particularly difficult to predict and which occurs preferntially during La Nina events. Two other clusters found in many winters have no observational counterparts. A regime which is very similar to the seasonal mean response to cold tropical Pacific SSTs is seen during several La Nina winters.

A strong negative correlation between a measure of the strength of the clustering and the Nino3 SST index is found. That this correlation is as strong as the correlation between the seasonal mean response and the same SST index indicates that the El Nino-Southern Oscillation related SSTs affect the regime structure of intra-seasonal flow as strongly as they do the mean state.

Recently there have been high level discussions within WCRP, which I have attended, concerning future scientific directions and challenges within WCRP. A key concern has been to ensure that WCRP projects are relevant to the current scientific environment and that WCRP is able to capitalise on its achievements so far and remain at the forefront in the field, particularly in predictive activities. Furthermore it is recognised that WCRP needs to position itself to better meet the challenge posed by other organisations and to ensure that its projects are attractive to younger scientists. As a result of these discussions the concept of a WCRP Climate system Observation and Prediction Experiment (COPE) has emerged. The rationale, objectives and a tentative structure of COPE will be presented.

The NSSL Warning Decision Support System - Integrated Information (WDSS-II) has provided an invaluable development environment to facilitate the development of these new applications. Since 2002, NSSL created new, and improved multiple-sensor algorithms. NSSL has converted its suite of single-radar severe weather detection algorithms to operate using multiple radars. NSSL has also developed a suite of new radar diagnostic derivatives, including two-dimensional high-resolution fields of vertically-integrated liquid (VIL), VIL Density, Probability of Severe Hail, Maximum Expected Hail Size, Velocity-Derived Rotation, and Velocity-Derived Divergence. Time-integrated fields of some of the above have also been developed, including hailswath information (maximum size and hail damage potential) and velocity-derived rotation tracks. Other new applications include improved multiple-sensor data quality control applications, a two-dimensional reflectivity motion estimation algorithm, and applications designed to blend data from WSR-88D and TDWR radars. We have evaluated the operational utility of the new severe weather algorithms (and the new display concepts) in real-time NWSFO proof-of-concept test settings. We will describe the new algorithms and explain how the WDSSII facilitated the development process of the algorithms. These new concepts will continue to be tested to determine whether they will be included in future operational National Weather Service systems that help guide and manage the severe weather warning decision making process.

Several conclusions can be drawn. First, while the cloud regimes have been derived for the entire TWP region, data at the ARM sites only can be used to consistently characterize the regime properties. This indicates that the ARM sites are representative sites to measure the key features encountered in TWP cloud systems. This result is of further significance, since it might enable the use of measurements only obtainable at those sites (e.g., vertical profiles of cloud fraction) to characterize cloud properties over the entire TWP. The potential benefit of cloud regime dependent analysis of model errors for the development of parameterizations in GCMs will also be highlighted.

An extended 50-level version of LAPS, with the model top raised to 0.1 hPa and nested within a similarly extended GASP, is being used to test the 1DVAR assimilation of locally received ATOVS radiances, processed via the AAPP package. The timeliness of local reception and processing should improve the amount of ATOVS data available to the LAPS system which, operationally, employs an early data cut-off. The vertical extension of both models eliminates the need for NESDIS retrievals and promises a fully unified local/global data assimilation system able to handle radiance data, whether received and processed locally or sent from overseas centres via the GTS, equivalently.

CEOP aims to provide data from a variety of sources (field data, model data from major NWP centres and relevant satellite data) in common formats with easy access to all interested parties. Hence one component of CEOP is investigating the best way to hold the data and techniques to allow value-added access to it. Other subprojects include the WESP (Water and Energy Simulation and Prediction) group chaired by John Roads and the CIMS (CEOP Inter-monsoon Modeling Studies) group chaired by Bill Lau.

I recently attended the First CEOP Model Output Development and Analysis Workshop and the following 3rd International Implementation Planning Meeting. The seminar will describe CEOP, its rise and rise, the databases and available data and discuss a few of the issues raised at the meetings. I will also discuss the current status of the data and the contribution BMRC is making to the project.

2/ Statistical forecasting TC track model:
MOCCANA is a statistical model which is similar to CLIPER model. This model will be presented and also the comparison with CLIPER will also be given.

This presentation describes a project designed to help the people living along the coast of the Beaufort and Chukchi Seas adapt to climate change and variability on a more informed basis. This talk will review our investigations of cyclone climatology and dynamics; how we have organized to work across disciplines and with the community; and community policies that have been informed by our work so far. We have learned that human and environmental factors combine in unique way in this community; in other words, context matters in adaptation to climate change and variability.

In collaboration with colleagues from the Department of Aerospace Engineering, the Institute for Arctic and Alpine Research and the Center for Public Policy Research at the University of Colorado, the National Center for Atmospheric Research, the Barrow Arctic Science Consortium, and the Georgia Institute of Technology

* Note this seminar will be given in the 5th floor conference room in conjunction with an AMOS function.

The goal of this study was to develop a statistical scheme to predict the probability of genesis and occurrence of TCs during subsequent weeks, over several areas in the Southern Hemisphere. The chosen statistical model is lagged multiple logistic regression. Besides just the MJO, other large-scale influences were also investigated: phase of ENSO, position within the cyclonic season, phase of the QBO, etc. In this seminar, the cross-validated performance of the developed model will be shown.

To consistent with the operational fog forecasting process at WA Regional Forecasting Center, this NWP fog guidance is risk categories forecasting. Considering the uncertainty of the model outputs and natural difficulty of fog forecasting, fuzzy logical method has been employed to develop the NWP guidance. The fuzzy functions are based on both the fog climate and Meso_LAPS model analysis. The fog guidance has been used at WA Regional Forecasting Center on year 2003 fog season and the verification results are discussed.

This seminar will describe my recent work on extending Single Column Modelling to the Tropical Western Pacific (TWP) ARM sites at Nauru and Manus Island. Several SCM forcing data sets have been derived for the TWP from global NWP data sets. This has enabled us to run ensemble SCM simulations using both the ECMWF and BAM single column models at Manus Island and Nauru.

The presentation will cover:

• A brief introduction on Single Column Models.
• Derivation and verification of SCM forcings in the Tropical Western Pacific.
• Initial results from ensemble runs using the ECMWF SCM.

There will also be discussion of the enhanced diagnostics for monitoring observations and model performance that will soon be available as well as an outline of future developments.

This seminar is skill-score free!

This talk examines the influence of Bureau information on the portrayal of long-term rainfall outlooks in selected print media during June to November 2002. The analysis period corresponds to when the El Niño cycle has its greatest influence on Australian rainfall. Rainfall outlooks from the Bureau are at their most accurate during this time, and Australia is most likely to be suffering from low rainfall and (possible) resultant drought.

Content analysis was undertaken on drought-related articles from four Victorian newspapers. Of particular interest were those articles that cited the Bureau rainfall outlooks, and their general presentation. From the articles identified, a journalist from each newspaper was interviewed to investigate their use of rainfall outlook information.

Overall, Bureau rainfall outlook information was well transmitted for the period and papers studied. The Bureau was attributed as the source of the information in 46% of outlook articles recovered. The accuracy of information published was very high (87%). Only 20% of articles directly related to the Bureau's monthly media releases. However, information on the web and discussions with Bureau staff seemed to provide adequate content for journalist's needs between release dates. Nonetheless, further improvements could be made to this communication process.

Clare Mullen is a senior meteorologist with the Bureau. This talk stems from a minor thesis undertaken towards a Masters of Arts (Communications) at RMIT, completed in 2003.

Opportunities to improve QPF in landfalling TCs seem particularly promising because of improved understanding of TC precipitation mechanisms, quantitative precipitation estimation (QPE) from remote sensors (both ground and space-based), and improvements in horizontal resolution of operational and research models so that moist non-hydrostatic processes, which occur on small scales, may be represented better. However, a major stumbling block to improved QPF in TCs is a lack of a comprehensive climatology of TC precipitation, a description of the distribution of rain in space and time, using any measurement technique. Few precipitation climatologies exist for TCs in the United States, and other TC basins have similarly limited climatologies. New QPE opportunities using satellite based QPE techniques (e.g, TRMM, AMSU, AMSR) offer a unique opportunity to develop TC rain climatologies for the US and globally. These climatologies are being used as a baseline to validate numerical model rain forecasts, and thereby improve QPF in TCs. Opportunities are identified and some sample comparisons presented.

The unique chain of events starts in the South China Sea, with the monsoons blowing in the direction of its length (from south-west in summer and north-east in winter). In summer the wind driven warm surface layer veers to the right (eastwards) under Coriolis forcing (Ekman drift), piling up against the coasts of Borneo and the Philippines. The winter monsoon from the Asian continent cools the surface water, which sinks through the warmer water below. This destabilizes the thermal stratification and generates vertical turbulence, impeding the drift of the surface layer. The overall result is an accumulation of warm water in an elongated pool, leaning against the coasts of Borneo, Palawan and later Luzon. The pool will grow over several years with moderate cooling during the winters. As its depth increases, thermal expansion raises its surface above the equipotential level. The resulting horizontal pressure gradient generates an outflow, which veers to the right under Coriolis forcing until the flow is at right angle to the pressure gradient. This creates a geostrophic flow towards the north-east along the western boundary of the pool. In contrast to oceanic gyres with anticyclonic flow around their warm core, this pool has a boundary current only in the west, reducing it to a half-gyre. Once it is deep enough, its boundary current will strengthen and erode it from the south, making the pool migrate to the north. When its front reaches the northern tip of Luzon, Coriolis force makes the boundary current turn to the right and overflow into the Pacific Ocean.

The Kuroshio begins where the ocean spanning North Equatorial Current arrives before the east coast of the Philippines. Both are anticyclonic currents around the warm core of the Pacific subtropical gyre and have cool depth water to their left. In our hypothesis the warm water overflowing from the South China Sea replaces the cool depth water between the Philippines and the Kuroshio, destroying the pressure gradient needed for geostrophic flow. As a consequence the Coriolis force makes the Kuroshio veer sharply to the right, breaking it up into anticyclonic eddies. This leaves the waters arriving with the North Equatorial Current stranded off the east coast of the Philippines, where they accumulate and deepen the warm pool in the western Pacific. It is not possible that the stranded warm waters can restore the Kuroshio because there is no negative pressure gradient between them and the coast, necessary for a geostrophic northward flow. The only way in which the Pacific gyre can recover is through a removal of the stranded waters. Data from satellite altimetry, showing the sea-surface topography resulting from thermal expansion, indicate that in December 1996 the warm pool between the Philippines and the Marianas shifted south-eastwards. Subsequently it remained almost stationary, stretching along the equator with its center slightly west of the dateline, until mid March 1997. Then the warm water surged eastwards as the record 1997-98 El Nino evolved.

The presentation will briefly review the impacts of volcanic ash on aviation, examine some of the uncertainties and describe current activities that are directed at improving the use of satellite data and the dispersion model output and streamlining the warning preparation process.

Please register your attendance by contacting Ruth Byrne via email at r.byrne@bom.gov.au, or by phone on ext 4384.

Hourly Numerical Weather Prediction (NWP) model forecast errors are investigated, demonstrating that their nature varies with a strong diurnal modulation overlaying the normal error growth with time. The OCF algorithm used for daily forecast fields is shown to perform well for hourly forecasts where the bias correction and weighting strategy is applied to each individual forecast hour. A method by which a "blended" hourly OCF forecast can be achieved using a mix of model forecast intervals is described and is shown to increase the accuracy of hourly OCF forecasts.

Besides having an unusual grid, C-CAM possesses a number of attractive features in its dynamical formulation. Perhaps the most important of these is a reversible staggering procedure for the winds, available because of the cyclic nature of the grid. This procedure provides very similar dispersion properties to a spectral or vorticity/divergence formulation.

C-CAM is a two-time-level, semi-Lagrangian, semi-implicit model, allowing large advective time steps. For vertical advection, the TVD scheme is found to represent the tropopause best.

C-CAM has some features that can probably only be used in a grid-point model. One is a self-consistent treatment of pressure-gradient terms near orography, which is possible because of the reversible staggering procedure. Another is an "exact" treatment of the orographic contribution to the semi-Lagrangian advection of surface pressure. A new treatment of temperature advection near orography also obviates most of the need for hybrid coordinates. It has also been found possible to eliminate the usual semi-Lagragian off-centering in time over most of the domain, thereby leading to improved tropical rainfall.

Some other features of C-CAM will be briefly described. One is its a posteriori scheme for conservation of mass and moisture. Another is the use of the Schmidt transformation to provide smoothly-changing variable resolution. A nonhydrostatic version of the model will be briefly described, and also an outline of the MPI architecture implemented in the model.

A recent realization is that the various dynamics enhancements developed for C-CAM can be applied without great difficulty on a non-orthogonal gnomonic-cubic grid. This new atmospheric model will be known as G-CAM. Its attractions will be great uniformity of resolution, and straightforward incorporation of highly-accurate fully-conserving advection for trace gases and chemical species.

Our approach is to develop some idealised systems of small dimension, in which algorithms can be readily tested and compared. This testing has included the differences between perturbed-observation (PO) and non-PO schemes, the impact of nonlinearity, covariance inflation, single vs double filters, covariance localisation and regularisation, and the use of an ensemble representation of the observation-error covariance. Results from a (hopefully) interesting and instructive subset of these will be presented.

We analyse how well these various vertical grids simulate the general circulation and deal with the vertical-grid-scale noise. Our method to study the latter is based on the approach of Willamson et al. (1998) where they isolate vertical-grid-scale noise by applying a 1-2-1 filter in the vertical to a simulated model field.

From this study a set of rules is formulated which may be usefull in selecting the location of vertical levels in GCMs.

Significant progress has been made in land-surface observation and modeling at a wide range of scales. Projects such as the International Satellite Land Surface Climatology Project (ISLSCP), the Global Soil Wetness Project (GSWP), and the GEWEX Continental-Scale International Project (GCIP), among others have paved the way for the development of an operational LDAS. Several LDAS systems have been implemented in near real time and at high spatial resolution for North American, European, and global domains. These LDASs are forced with real time output from numerical prediction models, satellite data, and radar precipitation measurements, and can incorporate land state observations as a constraint to the model dynamics using hydrologic data assimilation methods. Results of LDAS assimilation of land surface temperature, moisture, and snow are showing great promise to improve predictability and understanding of model realism.

A new version of the system (POAMA-2) is currently being development and will be implemented operationally in 2005. For POAMA-2 a large ensemble set of hind-casts covering the last 25 years will be produced. One of the new focuses for POAMA-2 is the prediction of rainfall in the Australian/Asian region. An overview of the POAMA-2 system will be given.

First, the impact of neglecting surface ocean currents in the calculation of surface stress is assessed by conducting an experiment in which they are taken into account. Compared to a control experiment without the shear correction there are several significant improvements: The model's equatorial SST cold bias is reduced by over 1 degree Celsius in the western central Pacific. There is a reduction of equatorial interannual variability of upto 30% to more realistic levels. The dominant period also increases to around 4 years, and the propagation characteristics of SST anomalies change from westward to an almost standing pattern.

Second, the impact of neglecting temporal and spatial variations in optical penetration depth due to ocean biological processes is assessed by performing an experiment in which the HAMOCC5 biogeochemistry model is used in the climate model to estimate these variations. Compared to the simulation with wind shear correction there are several further improvements: The equatorial cold bias in the eastern Pacific is reduced by upto 0.4 degree Celsius. The phase of the simulated annual cycle is shifter earlier by about two weeks, improving the simulation. Interannual SST variability weakens by a further 10-15%, to even more realistic levels. The dominant period is lengthened further to around 5 years.

In both cases, the reasons for changes in ENSO characteristics are due to both changes in the mean state and to changes in the relevant feedbacks. These shall be discussed.

The experiment involved verifying a set of subjectively derived quantitative forecasts for Melbourne (Australia) out to 14 days. The verification data suggested that, at that time, routinely providing or utilising day-to-day forecasts beyond day 4 would be inappropriate, but it might have been possible to provide some useful information about the likely weather up to about a week in advance for some elements and in some situations. By contrast, the data also suggested that in some circumstances it may not have been possible to provide useful information even for days 3 and 4. Shortly thereafter, in April 1998, the Victorian Regional Forecasting Centre (RFC) commenced a formal trial of forecasts for Melbourne out to 7 days.

There have been considerable advances in Numerical Weather Prediction modelling since then, and also in associated techniques for statistically interpreting the NWP model output utilising objective methods.

In July 2004, the current author presented an analysis of results of the trial forecasts of maximum temperature to the Australian Meteorology and Oceanography Society 2004 Annual Conference and showed that over the period of the trial from 1998 to 2003 that forecasting skill was displayed out to 7 days. At about this time, RFC forecaster Stuart Coombs alerted the author to anecdotal evidence that the output of the NOAA GFS Long Range (384-Hour/16 Day) NWP Model displayed considerable skill, and that, on occasions, it had predicted significant events even towards the end of the forecast period.

The purpose of the present paper is to utilise a knowledge based forecasting system to objectively interpret the output of this Long Range Model statistically in terms of local weather:

• maximum temperature,
• minimum temperature,
• probability of precipitation, and
• amount of precipitation

• there is some predictability 8 days ahead and beyond,
• most of this predictability is associated with the forecasting of temperature (especially maximum temperature),
• the probability that this encouraging result has arisen "by chance" is 2.63%,
• an ongoing trial is warranted.