Editorial Hello, and welcome to the April issue of the Bulletin of the Australian Meteorological and Oceanographic Society (BAMOS). My name is Linden Ashcroft, and it is with joy that I introduce myself as your new editor. It is very exciting for me to be able to write this editorial, as I have recently returned to the world of oceanic and atmospheric sciences. I completed my undergraduate studies in meteorology at the University of Melbourne, with an Honours year in 2007, and worked as a research assistant in 2008. However, in 2009 however I left the nation’s cultural capital for its actual capital to undertake a Graduate Diploma in Science Communication at the Australian National University. This unique course, run in conjunction with Questacon – the National Science and Technology Centre in Canberra, opened my eyes to the growing world of science communication. I learnt how to make slime with school children, how to write scientific articles and how to share science with anyone from pre-schoolers to policy makers. With these new skills in hand, and a renewed love not only of climate science but the crucial importance of sharing it, I returned to the University of Melbourne last month to begin a PhD under the supervision of Joëlle Gergis and David Karoly. My research will focus on reconstructing the south east Australian climate from 1788 using early instrumental data.
Firstly, I have had help. The previous editor, Stewart Allen and I have worked on this edition together, and it is thanks to his tireless enthusiasm and dedication that I am here at all. He has made the transition an easy process, and has nurtured the Bulletin constantly over the past two years. I thank him for all of his efforts and personally express my gratitude for his assistance with this issue. Secondly, I am passionate about sharing science with everyone and I am excited that I have the opportunity to put this passion to good use. Communicating quality science about our planet is vitally important, and is something that as a scientific community we need to embrace. In this issue you will find a very interesting conference report by Joëlle Gergis and Ailie Gallant on their experiences at the recent Science meets Parliament program that is a testament to this. With that in mind, I strongly encourage you to continue sharing your stories here. Keep your community up to date on the science, issues and events that are important to you, by way of reports, articles, centre updates, photos, or simply a note about an interesting website or book. Whatever it is, I want to hear about it, and I hope to continue the Bulletin’s role in keeping all members informed. I have a few new ideas slowly sprouting for the Bulletin, and I would also very much like to hear yours. Until then, enjoy this issue of BAMOS.
Although I am new to this field, and indeed to AMOS, I feel excitement rather than fear as I step into this role as your editor. This is for two main reasons.
Bulletin of the Australian Meteorological and Oceanographic Society Vol. 23 page 21
Linden Ashcroft
President’s Column Global warming continues I have been looking at the global temperature record, as measured by satellite and surface data, both of which are readily available. I have plotted the time series of six-month (September–February) global mean temperature anomalies in the figure below. The data I have used are the Spencer-Christy lower tropospheric temperatures from satellites2 (labelled “UAH” in the figure) and the surface temperature data from the NASA Goddard Institute for Space Studies3 (labelled “GISS). Both data sets are of anomalies from the period 1979/80-1998/99, i.e. the first twenty years of the satellite observations. The GISS data originally used a different base period, so I have recalculated the GISS time-series with the same base period as the UAH data. Please look at the data, re-do the analysis, and see if you come to similar conclusions to those I describe below.
data sets. Even the interannual variations in the two temperature series are close matches. A corollary of this close match between the surface and satellite variations and trends is that the surface warming trend is not due to the urban heat island effect. Nor is it due to decreases in the numbers of stations used in the surface analysis. Neither of these problems would affect the satellite data. Furthermore, the trend is not simply due to a year or two of extreme temperatures – the very warm years (often associated with an El Niño) are progressively getting hotter, as are the cool years (which are often associated with a La Niña). Another corollary is that the GISS data provide a very credible analysis of variations and changes in global mean temperature (since they match the satellite observations), supporting the conclusion that the 21st century has seen the warmest September–February
Figure 1: Global September-February temperature anomalies. Surface data (GISS) is shown in grey, satellite data is shown in black. Anomalies are relative to 1979/80 – 1998/9. The last six months (September 2009 – February 2010) have been the warmest September-February observed in the satellite record, by a large margin. In the GISS surface data the last six months are the equal warmest September–February on record (equal with 2007/08). Eyeballing the graphs of the surface and satellite temperature record should convince anyone that global warming continues unabated. Fitting a linear trend to the data since the start of the satellite observations produces virtually identical trends in the two
2
vortex.nsstc.uah.edu/data/msu/t2lt/uahncdc.lt
3
data.giss.nasa.gov/gistemp/tabledata/GLB.Ts+dSST.txt
periods for at least 130 years. As I write this column (15 March) the Spencer-Christy satellite data for March 2010 are running well above the previous March record (set in 2004). November 2009 was the hottest November in the dataset and January 2010 was the hottest January, with February 2010 coming in as the second hottest February. It doesn’t look like the run of record hot global temperatures will stop anytime soon. Neville Nicholls (This is an edited version of the President’s Column available at www.amos.org.au - Ed.)
Bulletin of the Australian Meteorological and Oceanographic Society Vol. 23 page 22
News Observations for weather and climate in the Southern Ocean Eric Schulz The Centre for Australian Weather and Climate Research, Bureau of Meteorology, Melbourne For the first time, a weather buoy has been deployed in the remote Southern Ocean. It is moored in water 4.6km deep, 580km southwest of Tasmania at 46.75S, 142E. Since March 2010, the Southern Ocean Flux Station (SOFS) has been relaying hourly observations of the wind, temperature, humidity, air pressure, sunlight and rain back to shore, giving insight into the current conditions “down south” as well as helping to build a record in this climatically important region of the global ocean. As well as aiding weather forecasting, the meteorological observations are used to compute the total heat, mass and momentum fluxes on time scales from diurnal up to annual. While the Southern Ocean plays a significant role in the global climate system, there is a paucity of sustained in situ air-sea flux observations in this harsh and remote region. This lack of “ground truth” has led to significant differences in global flux products in the mid to high latitudes. The high quality observations are a valuable contribution to building the climate record and understanding climate variability. SOFS is operated by the Bureau of Meteorology and is one of five platforms deployed as part of the Southern Ocean Time Series (SOTS): a multidisciplinary ocean observatory at the Subantarctic Zone operated by the Bureau, CSIRO and the University of Tasmania. The other platforms are gliders, autonomous profiling floats, sediment traps and the Pulse bio-geo-chemical moorings. SOTS is tasked with collecting sustained observations of the atmospheric surface layer, as well as the upper and deep ocean to understand the transfer of heat, moisture, energy and carbon dioxide between the atmosphere and ocean. This will improve our knowledge of climate, carbon processes and the role of the oceanic ecosystem. SOTS is one of a few high temporal resolution time-series sites identified in the OceanSITES project and one of three proposed for the Southern Ocean.
The SOFS mooring “slack line” design (not to scale) displaying the distribution of sensors. The mooring length is 1.33 times the water depth to reduce line tension.
SOTS is a facility of the Australian Integrated Marine Observing System (IMOS). IMOS is funded through the National Collaborative Research Infrastructure Strategy (NCRIS) and
Bulletin of the Australian Meteorological and Oceanographic Society Vol. 23 page 23
the Education Investment Fund (EIF) to deliver data streams from the oceans around Australia. IMOS aims to meet the needs of the marine and climate research community, address issues of national importance and contribute to international ocean observing programs.
Further information:
The near-real-time observations from SOFS and other stations are available from the IMOS Ocean Portal.
Ocean Portal: imos.aodn.org.au/webportal
SOFS: imos.org.au/sofs.html SOTS: imos.org.au/abos.html OceanSITES: www.oceansites.org IMOS: imos.org.au
Disputed isle in Bay of Bengal disappears into sea In late March, the global media was reporting that, according to Indian scientists, an uninhabited island in the Bay of Bengal had vanished. New Moore Island, located south of the Hariabhanga river, had been claimed by both Bangladesh and India for almost thirty years, since it first appeared in the seventies. However, oceanographer Sugata Hazra, a professor at Jadavpur University in Calcutta, noted the island’s disappearance on satellite imagery and later confirmed it with fishermen and sea patrols. India and Bangladesh both claimed the empty New Moore Island, which was about 3.5 kilometers (2 miles) long and 3 kilometers (1.5 miles) wide. Bangladesh referred to the island as South Talpatti. Another nearby island, Lohachara, was submerged in 1996, forcing its 4000
inhabitants to move to the mainland, while almost half the land of Ghoramara Island was underwater, Professor Hazra said. He also stated that at least 10 other islands in the area were at risk as well. Hazra attributed the island's disappearance to global warming, saying: "Coastal erosion and rising temperatures in the Bay of Bengal between 2000 and 2009 led to New Moore Island getting submerged." Hazra commented that until 2000, sea levels in the area increased approximately three millimetres annually, but in the past year the figure was augmented to about five millimetres. Further information: timesofindia.indiatimes.com/home/ environment/global-warming/NewMoore-isle-no-more-expert-blameswarming/articleshow/5720685.cms
Lake Eyre floods - again Recent rainfall and flooding in Queensland mean that Lake Eyre will flood for the second year running. However, this might not be enough for the Lake Eyre Yacht Club to hold its 2010 Regatta. Initially, the yacht club suggested that flood levels would be the highest since 1989-90, bringing excellent sailing conditions – in a strictly locally relative sense – for their 10-year anniversary. Recent updates though have dried out these hopes. The last status report asserts that while the lake will look good from the air, it will ‘probably not be navigable’. The levels in the lake have historically been deduced and recorded using mostly anecdotal
observations. Given the scale of the lake – a 3000km round trip – this was the only available method for many years, however it led to wildly conflicting reports. A more accurate method of using calibrated satellite imagery is now used and direct observations of lake depth appear to verify the accuracy of these measurements to within 100mm in the navigation critical 1 to 3 metre depth range. Although the Lake Eyre Yacht Club is only 10 years old, boating on the lake during times of flood has been occurring since at least 1950. Further information: www.LakeEyreYC.com/Status/ latest.html
Bulletin of the Australian Meteorological and Oceanographic Society Vol. 23 page 24
Conference Report Communicating climate change: advice from Science meets Parliament 2010 Joëlle Gergis and Ailie Gallant School of Earth Sciences, University of Melbourne If you had the attention of a politician for five minutes, what would you say? How would you summarise the work that you do, or the importance of acting on climate change in less than 45 seconds – the time it takes for a sparkler to burn itself out? This is exactly the dilemma 120 early to midcareer scientists faced during the 11th Science Meets Parliament (SmP) held in Canberra on 9–10 March 2010. Australia’s leading science advocacy organisation, the Federation of Australian Scientific and Technological Societies (FASTS), runs SmP every year to provide an opportunity for Australian scientists to be exposed to the political process and how science can influence decision makers. This year AMOS generously supported Dr Ailie Gallant and Dr Joëlle Gergis, research fellows from the School of Earth Sciences at the University of Melbourne, to attend. Day 1 was a series of professional development seminars designed to provide us with a glimpse of how the media, policy development and effective communication actually work. We heard from Kevin Rudd’s speechwriter Tim Dixon, Alison Carabine from ABC’s Radio National and Richard Dennis, executive director of the Australia Institute. We discussed the different cultures that journalists and politicians inhabit and some of the barriers to having our science understood. We learned the importance of ‘knowing your audience’ and targeting your message with the right language and within the right context. We were told that when it comes to communicating complexity, it’s always best to try and humanise the scale of what you are trying to say. People need to know how your information affects their daily lives (or the prosperity of the nation); before you can compel someone to take action, they need to very clearly understand the risk of inaction. It was in this context that we discussed climate change, the unofficial theme of this year’s SmP. Following the journalists’ admission that conflict makes a good story, we discussed the ethics of providing a voice to global warming contrarians in the name of journalistic ‘balance’. They explained that the public is still trying to assimilate the complexity of climate science and very often do not possess
the critical thinking to distinguish the weight of opinion filtered through the peer-reviewed literature and opinions espoused through nonspecialists in the blogosphere. When a controversial view on climate change crops up, the journalists admit that the media seize it as a ‘fresh angle’ on a now long running story that is starting to sound like more of the same to the general public. As Herald Sun columnist Andrew Bolt understands, controversy will always draw a crowd. Lively discussions buzzed into the evening as we made our way into the Gala dinner in the Great Hall at Parliament House. After a welcome by Kim Carr, the Federal Minister for Innovation, Industry, Science and Research, SmP guests wined and dined on tables sprinkled with the likes of Minister Lindsay Tanner, Senator Steve Fielding and the deputy leader of the opposition, Julie Bishop. ABC broadcaster Robyn Williams hosted us through an entertaining and thought provoking evening of special guests. The keynote speaker, Chair of the Australian Science Media Centre, Mr Peter Yates, made the perceptive comment that the planet’s epitaph might read: ‘We got the science right but we stuffed up the communication’. He cautioned the climate community’s recent trend of avoiding addressing the arguments of extremely vocal, well-orchestrated global warming contrarians. He even went as far as to suggest that climate science needs a key spokesperson to do what Carl Sagan did to lift the profile and popular understanding of the complex field of
This year’s AMOS SmP attendees, Joëlle Gergis (L) and Ailie Gallant (R) (Photo: A. Gallant).
Bulletin of the Australian Meteorological and Oceanographic Society Vol. 23 page 25
astronomy. Discussion suggested that the recent development of Climate Scientists Australia, an independent group of our senior scientists willing to provide evidence-based information to climate related policy decisions, could fill the vacuum created by the collapse of discussions surrounding the Carbon Pollution Reduction Scheme (CPRS) and international negotiations at Copenhagen last year. It was reassuring to meet with parliamentarians who had an appreciation of climate science and to hear examples of infiltration of science into policy making. A specific example of this came from a meeting between Ailie Gallant and ALP backbencher, Mike Symon, which highlighted that societies such as AMOS can influence government policy. In his capacity on the committee for Innovation, Industry, Science and Research, Mr. Symon demonstrated that he had a good grasp of an inquiry into long-term meteorological forecasting in Australia and specifically mentioned the difficulties faced in forecasting on time scales from years to decades. This was an inquiry to which AMOS had made a submission and several points made by AMOS were included in the final report. A second meeting between Ailie and an MP from Queensland, Yvette D’Ath from the marginal seat of Petrie (north of Brisbane) highlighted the community responses to the recent media attention on mistakes in the IPCC report and the “Climategate” email debacle. Ms. D’Ath expressed concern that many in her electorate who had only recently come to accept that humans have had a discernable influence on the climate system now seriously doubted these claims. Ms. D’Ath was very interested to learn ways in which she could respond to those in her electorate claiming the science was incorrect and asked for websites and other sources that she could pass on to people in her electorate. Joëlle Gergis met with Mr Petro Georgiou, a liberal holding the seat of Kooyong in Melbourne’s affluent eastern suburbs. As a man who crossed the floor in support of the CPRS Bill in November 2009, he needed no convincing about the dangers climate change poses to our economy and lifestyle. He believed that one of the key mistakes made by the Rudd government was trying to rush through the complexity of decarbonising the Australian economy in step with the Copenhagen deadline. He felt that the public and parliamentarians did not really understand the CPRS so were not prepared to compromise the status quo.
On day 2 we were treated to a fantastic guest speaker, American science writer Chris Mooney, at the National Press Club. He gave an incisive overview of the nature of the ‘guerrilla war’ being waged on climate science in the untamed jungles of the online world. He said it was naïve for scientist to feel that the ‘truth will prevail’ in the global warming debate as the mountain of peer-reviewed evidence grows. Instead he suggested that as a community we need to equip ourselves with the professional communication skills needed to combat the very targeted tactics of our opponents. In a recent interview Professor Michael Mann (co-creator of the ‘hockey stick’ temperature reconstruction) referred to the ‘asymmetric warfare’ between trained global warming contrarians and climate scientists as ‘literally like a battle between a Marine and a Cub Scout’. In the 11 March 2010 issue of Nature, the editor warned that ‘scientists must acknowledge that they are in a street fight, and that their relationship with the media really matters’. Chris Mooney suggested that climate scientists simply have not received the core communication training they need to fight the war. He proposed that we must begin to train a small army of ambassadors who can translate the science and make it relevant to the media, politicians and the public. It was inspiring to hear that short science communication courses are now being offered to students at the University of California’s Scripps Institution of Oceanography, with further plans to extend this to Princeton University later this year. No doubt these courses aimed at training ‘bridge builders’ of the future will help 21st century scientist harness the enormous influence of the online world in a constructive – rather than destructive – way. At the end of our time in Canberra, we left with the clear message that scientists are welcome in the political process, but we must equip ourselves with the tools of effective communication our knowledge is to be heard. We need to be prepared to defend our science in the face of intense public scrutiny with conviction and in plain English. We learnt that, if possible, we need to tell a human story and to say something new, while remembering to talk with the audience, and not at them. Once we restore community confidence in climate science, one conversation at a time, our politicians will have no choice but to follow. Here’s that sparkler; your time starts now. Further Information: www.fasts.org
Bulletin of the Australian Meteorological and Oceanographic Society Vol. 23 page 26
News from the Centres Presentation of 2009 Priestley Medal AMOS president Neville Nicholls visited Hobart on the 12th of March to present the 2009 Priestley Medal to Dr Susan Wijffels of CSIRO, at an AMOS function in the CSIRO Marine Laboratories. The Priestley Medal is awarded to a younger scientist who has achieved excellence in Australian meteorological, oceanographic or climate research and is presented every second year.
catalyst to reinvigorate the Hobart Centre, and encourage other regional centres to propose similar events in the future. Presentation of AMOS prizes is an important part of the Society’s role, and recipients deserve to be celebrated.
About 50 AMOS members, colleagues and members of Susan’s family participated in the function. Dr Trevor McDougall opened proceedings with a brief overview of Susan’s scientific contributions before Neville presented the Medal to Susan and gave a short presentation about AMOS and how the National Council sees the Society developing. Susan then provided an excellent presentation on her recent work. Mr Paul Durack, who is completing a PhD under Susan’s supervisor, discussed their work together. Afterwards, everyone moved to the CSIRO canteen for a “happy hour”, with drinks and nibbles provided by the AMOS National Council. Trevor McDougall, Paul Durack, and Dr Mike Pook did a great job of arranging the function. It is hoped that this event will be the
President Neville Nicholls presents the 2009 Priestley Medal to Dr Susan Wijffels of CSIRO in Hobart (Photo: T. Moore).
AMOS Weather Tipping Vaughan Barras The Centre for Australian Weather and Climate Research, Bureau of Meteorology, Melbourne For the past 5 years the Melbourne Centre has run an annual AMOS Weather Tipping Competition. This is an online weather forecasting competition that operates along the lines of the familiar football-tipping format, however instead of predicting winners, players are given the challenge of forecasting maximum and minimum temperatures and rainfall for selected locations around Australia and New Zealand. From humble beginnings the competition has now attracted a dedicated following not only from AMOS members but also members of the general public. The 6th year of the competition has seen a major redevelopment of the Weather Tipping website thanks to strong support from the AMOS National Council. This enabled the employment of Mr Michael Allen as web developer during the summer. He upgraded
the site comprehensively, adding new security features, user managed accounts and (for AMOS members) a graphic visualisation of forecast statistics for each round. The development of the website has been a collective effort, constantly receiving useful feedback from participants regarding the presentation of the site as well as many ‘headsup’ warnings which have saved me many headaches in my role as administrator. During the course of last season the administration of the site was handed over to PhD student Luke Garde at the University of Melbourne. I would like to acknowledge his dedication of many hours to this project, often out of his own time and it has been great to have him as part of the team. We have also received great support and numerous insights along the way from Blair Trewin, Grant Beard and Frank Woodcock at
Bulletin of the Australian Meteorological and Oceanographic Society Vol. 23 page 27
the Bureau of Meteorology and from Kevin Walsh, Brett Holman and Ailie Gallant at the University of Melbourne. We look forward to the excitement of this year’s competition that has attracted the highest number of participants we’ve ever had. We hope that all the players involved enjoy the
new website and have a bit of fun pitting their wits against fellow competitors! Further information, or to join: tipping.amos.org.au/ www.amos.org.au/news/id/65
Articles On emerging droughts Robert Fawcett, Blair Trewin and David Jones National Climate Centre, Bureau of Meteorology, Melbourne, Australia Address for correspondence: R Fawcett, National Climate Centre, Bureau of Meteorology, GPO Box 1289, Melbourne, Vic. 3001, Australia. Email: r.fawcett@bom.gov.au 1. Introduction Australia has unusually large rainfall variability due to its largely arid climate and its location on the edge of the western Pacific (e.g. Nicholls et al., 1997; Nicholls and Wong, 1990). Much of this large variability reflects the cycle of the El Niño-Southern Oscillation (ENSO) (e.g. Jones and Trewin, 2000) which is responsible for a regular cycling from drought to flooding rains. This substantial rainfall variability is important, as it is associated with frequent long-lasting and severe droughts with major economic, environmental and social consequences (McKeon et al., 2004). From a climate change perspective, large variability makes detecting underlying trends in rainfall statistically more difficult by decreasing the signal to noise ratio. This means that separating drought cycles (rainfall variations below a near-constant mean) from changes in underlying aridity (declines in the mean) may not be possible until some time after a trend becomes established or a step change to a drier climate has occurred. In a previous article (Fawcett, 2004), the time series of Melbourne annual rainfall was considered from the perspective of break-point methods, with evidence being presented that suggested a decline in rainfall since 1996. Similar results were obtained for grid-point rainfalls in the surrounding region. With the addition of four extra years of data, that result still stands, although in updating the calculation we take the opportunity to improve the procedure used to estimate the statistical significance of the result.
In the present article, we are interested in exploring how rapidly the statistical significance of the 1996/1997 break-point emerges with the addition of each subsequent year of data. Such a question has obvious relevance to water managers, for example, who must confront the question of when fluctuations in annual rainfall (such as apparent step changes) can no longer be considered just interannual variability but instead become indicative of climate change and/or variability on time scales longer than those covered by the available data. Such a statistical approach is complementary to physically based analyses such as presented by Timbal and Jones (2008), who linked rainfall declines in southeast Australia to changes in large-scale atmospheric fields (mean sea-level pressure and precipitable water). In this context, however, we note that if a particular site has rainfall records lasting a century for example, then a statistical analysis of the data (taken in isolation) cannot readily distinguish between a trend and oscillatory behaviour with a multi-centennial period. Therefore, since our approach is purely statistical, in this article, by “climate change”, we mean a transition to a new climate1 without reference to whether that change arises from an actual step change to a new climate regime or merely the exploration of a different part of a multi-centennial (or longer period) oscillation. Likewise, we do not attempt to attribute a cause to such a change. Following Fawcett (2004) our focus is southern Victoria, though we also perform an 1
That is, one beyond that indicated by the available observational data.
Bulletin of the Australian Meteorological and Oceanographic Society Vol. 23 page 28
analysis of rainfall in southwest Western Australia focussed on Perth and southeast Queensland focussed on Gatton near Brisbane. All three regions have experienced severe and protracted drought in recent decades (BoM 2008) and span quite different rainfall and climate regimes (BoM,, 2000). 2. Data For Melbourne station data, we use the time series of annual rainfall from the Melbourne Regional Office (RO) site (Bureau of Meteorology (BoM) station number 086071, location 37.807°S 144.970°E), supplemented by interpolations of three different sets of monthly gridded rainfall analyses. The first is the National Climate Centre’s operational monthly rainfall grid set (Jones and Weymouth, 1997), while the second and third are the new Australian Water Availability Project (AWAP) monthly rainfall grid sets (Jones, et al. 2007)2, analysed at 0.25° and 0.05° resolution respectively. These have been interpolated using bicubic polynomial interpolation3 to the Melbourne RO location. For Perth station data, we use the time series of annual rainfall from the Perth RO site (station number 009034, location 31.956°S 115.870°E). These data are supplemented by interpolations from gridded monthly rainfall analyses. For southeast Queensland, we use the Gatton site at the University of Queensland (station number 040082, location 27.544°S 152.337°E) as representative of the area hardest hit by recent low rainfalls (see Figure 4). We have chosen to use point data in these calculations, in order to obtain results comparable to those already reported (Fawcett 2004, 2005), but analogous calculations could obviously be performed on area-averaged rainfall data. 3. Analysis We begin by assessing the annual Melbourne station data (1856-2008) as a time series. As previously reported (Fawcett, 2004, 2005), the 2
AWAP rainfall maps and grids are available at www.bom.gov.au/cgi-bin/silo/reg/brs/rain_maps_awa.cgi. A technical report describing their construction is available at www.affashop.gov.au/PdfFiles/ awapfinalreport200710.pdfand a general description of the AWAP can be found at www.daffa.gov.au/brs/climateimpact/awap. 3
Conceptually, this involves fitting a polynomial p(x,y) =
data are approximately normally distributed with negligible autocorrelation (see also the Appendix). Cross-validation to minimise an error metric can be used to search amongst a range of different model types in order to explore the characteristics of the data. We have looked at five reasonably simple model types; constant (y = a), linear (y = a + bt), quadratic (y = a + bt + ct2), piecewise constant with a single break (y = y0 for t < t0 and y = y1 otherwise), and bilinear (y = y0 + a(t – t0) for t < t0 and y = y0 + b(t – t0) otherwise). For the first three model types, the best-fitting model is calculated via ordinary least-squares regression. For the last two model types, the fitting process involves the additional step of searching along the time series for the most appropriate value of t0. Mean square errors are computed under the leave-one-out single crossvalidation process, to circumvent problems arising from the models having different numbers of parameters to be estimated from the data. For this time series, the bilinear model provides the best fit (Figure 1), followed by the piecewise constant model. For the bilinear regression, the hinge point is at 1993 with an initial trend of 0.13 mm/year, followed by a subsequent trend of !14.0 mm/year (Figure 1). The confidence intervals for the regression and the data are also shown. These are estimated by means of a Monte Carlo simulation on the residuals (2001 iterations). The Monte Carlo simulation involves the initial calculation of the regression line and residuals, then resampling the residuals with replacement and adding them to the regression line to generate synthetic samples. Regressions are then calculated on the synthetic samples to construct the confidence intervals on the original regression line (blue lines in Figure 1). The confidence intervals on the data (green lines in Figure 1) are obtained by sampling the residuals of the original data (with replacement) and adding them to the regressions of the synthetic samples. As it seems rather implausible that such a rapid drying trend could persist long into the future, we also present the conceptually more plausible piece-wise constant results (Figure 2). The break is at 1996/1997, with the mean declining just over 20% from 660 mm to 520 mm. The statistical significance of this step change will be given in the following section.
to the data points (x,y,p) = (i,j,pij) for i,j = 0,1,2,3.
Bulletin of the Australian Meteorological and Oceanographic Society Vol. 23 page 29
Figure 1: Time series (1856-2008) of Melbourne RO (086071) annual rainfall in millimetres (red dots), together with the mean (light blue line), the best-fitting bilinear regression (solid black line), the 90% confidence interval for the location of the regression (blue lines) and the 90% confidence interval for the data (green lines).
Figure 2: Time series (1856-2008) of Melbourne RO (086071) annual rainfall in millimetres (red dots), together with the mean (light blue line), the best-fitting singlebreak piece-wise constant regression (solid black line), the 90% confidence interval for the location of the regression (blue lines) and the 90% confidence interval for the data (green lines).
In Figure 2, the last 12 years are all drier than the whole-period (1856-2008) mean (and consequently the 1856-1996 mean), a sequence not previously observed in the time series. In fact, the previous longest sequence of consecutive years below the whole-period mean is the six years 1979 –1984.
very much interested in actual rainfall values, and a new low annual rainfall record that broke the previous record by 100mm, for example, would be a much worse outcome than one that only broke the previous record by 10mm, a distinction not reflected in the ranks.
4. Melbourne – an emerging drought One way of exploring the emerging drought is to assess the null hypothesis of no change (in essence, that the annual rainfall time series comes from an identically independently distributed random process) at each year using only data available to that year and see how the statistical evidence leading to a rejection of the null hypothesis emerges over time with the arrival of new data. This allows a hindsight view of how rapidly decision makers might have been able to make policy adjustments in face of an apparent change in the rainfall climatology. Two different breakpoint methods have been explored. The first is the normal distribution parametric method described in Pettitt (1979) and Fawcett (2004), and the analogous nonparametric method of Pettitt (1979) which has the Mann-Whitney statistic as the key ingredient. The non-parametric method works off the ranks of the data, and does not assume any particular distributional form. In one sense, this is a decided advantage, as seasonal and annual rainfalls in Australia can exhibit strong departures (Clark and Brinkley, 2001) from the normality assumed in the parametric method. On the other hand, water managers are
The statistical significance of the results has been assessed in each case by means of a Monte Carlo simulation of 20,000 iterations involving resampling of the original sample to obtain an empirical sampling distribution for the relevant test statistic. The sampling was done with replacement for the parametric method, and without replacement for the nonparametric method. In the latter case, this was to avoid introducing tied values in the simulated samples that weren’t present in the original data. The Monte Carlo simulation is required, because theoretical calculations of significance (such as those used in Fawcett, 2004) typically assume a fixed pre-specified break location and don’t take into account the process of searching along the time series for the most plausible location for a possible break, thereby overstating the significance of any such break. For Figure 3, “Stn” denotes the Melbourne RO station data over the full period of 1856-2008, while “Short” denotes its restriction to the period 1900-2008 (to enable comparison with the grid-interpolated time series). “Opr” denotes the operational monthly rainfall grid sets (Jones and Weymouth, 1997), while “AWAP” and “HR” denote the AWAP monthly rainfall grid sets at 0.25° and 0.05°
Bulletin of the Australian Meteorological and Oceanographic Society Vol. 23 page 30
resolution respectively, all over the period 1900-2008. From 1998 onwards, 1996/1997 consistently emerges as the most likely place for a single downward step in the time series. Statistical significance, on the other hand, emerges more slowly. For the station data, it is attained in 2006 (p = 0.06 two-tailed parametric, 0.07 two-tailed non-parametric). The reason why the improvement in the significance occurs in 1998 rather than in 1997 in Figure 3 is because the break-point method as implemented in this study requires at least two years of data either side of the break. In other words, it takes ten years of the new (drier) Melbourne climate for it to become reasonably certain that the new climate is statistically different from the prior climate. The 1996/1997 break is clearly more significant in the station data than in the gridinterpolated data, but the reduced data availability for the latter is clearly not the cause of this difference. (The p-values associated with the linear trends in these data sets are not shown in Figure 3, because they do not attain significance.) A possible cause for the discrepancy could be data inhomogeneities in the station data and/or the gridded data. The Melbourne RO rainfall gauge moved three times between 1855 and 1863, and once more in 1908, but these moves
Figure 3: Breakpoint analysis on Melbourne RO (086071) annual rainfall. The horizontal axis shows the end year of the time series, while the vertical axis shows the statistical significance p value (one-tailed) for a single downward break somewhere in the time series. The significance is plotted on a logarithmic scale. “PBR” denotes the parametric breakpoint method and “NPBR” the non-parametric breakpoint method as described in the text. “Stn” and “Short” denote the station rainfall totals over 1856-2008 and 1900-2008 respectively. “Opr”, “AWAP” and “HR” denote interpolated grid rainfalls over 1900-2008.
were sufficiently early as to be unlikely to cause the discrepancy. The site is now heavily urbanised, but remains relatively unobstructed to the north, east and west with some high rise development to the south at a distance of approximately 50 metres. Record low 12-year (October 1996 to September 2008) rainfall around Melbourne covers a considerable area (Figure 4; see also BoM, 2008), and it would seem implausible that a substantial proportion of the rainfall stations in this area are experiencing data inhomogeneities resulting in artificial drying trends at the same time. Further, a comparison with a nearby site (Laverton Airport, station number 087031, location 37.856°S 144.757°E, approximately 19 km west of the Melbourne RO site) does not support the idea of a relatively recent data inhomogeneity at the Melbourne site (see the Appendix for more details). Additionally, we note that Timbal and Jones (2008) have found the rainfall decline in southeast Australia as being driven by largescale shifts in weather systems, further evidence that it is not artificial. Figure 5a shows time series of the ratio of the grid-interpolated annual rainfall to station rainfall for the three grid sets over the past 51 years, together with the linear trend over that period. There is a positive trend in the ratio for the two low-resolution analyses, but a slight negative trend in the high-resolution one. Over this period, the high-resolution analysis is the least biased of the three, which is not surprising given the strong climatological rainfall gradient over this region. The mean rainfall is substantially higher to the near east of Melbourne (peaking at near 1400 mm in the higher parts of the Dandenong Ranges 40 km east of the city) but only marginally less to the west (about 540 mm at Laverton), setting in place a strong and non-linear gradient across
Figure 4: Rainfall deficiencies for the 12year period October 1996 to September 2008, calculated using the AWAP lowresolution monthly rainfall analyses.
Bulletin of the Australian Meteorological and Oceanographic Society Vol. 23 page 31
Figure 5a: Ratio of grid-interpolated annual rainfall to station rainfall (1958-2008), Melbourne, together with linear trend lines.
Figure 5b: Ratio of grid-interpolated annual rainfall to station rainfall (1900-2008), together with linear trend lines.
the Melbourne RO site.
“Opr” interpolated grid rainfalls in Figure 6 for the post-1991 period.
Over the entire period of record, however (Figure 5b), all three grid sets show a wetting trend relative to the observations up to the 1960s, with little change subsequently. We note that the high resolution analysis products show very slight bias over the last 40 years which supports the conclusion that the Melbourne RO data are robust. The relatively slower rate at which the gridded product attains significance relates to these data having a slight dry bias at the Melbourne site during the first half of the twentieth century. It is not clear why this is the case, but is likely to relate to changes in the station network or the climatological gradients in the vicinity of Melbourne. 5. Perth It is instructive to compare the situation in Melbourne with another well-known example, that of Perth rainfall, where the rainfall decline has persisted since the 1970s (e.g., Nicholls et al., 1997). As before, we use grid-interpolated time series for the Perth RO site (009034). Additionally we use station data for this site, although it is only available for the period 1876 to 1991. Figure 6 shows the results for Perth, analogous to those for Melbourne presented in Figure 3, although in this case the assessment is carried out over the past 50 years. From 1970 onwards, 1968/1969 emerges consistently as the most plausible location for a single negative break, although early on occasionally interspersed with other years (1948, 1957, 1975). As previously noted, the annual 009034 station data cease in 1991. Patching the missing data with neighbouring stations 009151 (1992-1993, 7.2 km away) and 009225 (1994-2008, 4.0 km away) yields results (not shown) similar to those of the
Generally speaking, it appears that statistical significance emerged more slowly in the Perth annual rainfall time series than in the Melbourne data – significance had not been attained in the station data when the Perth station ceased reporting 23 years later. In the most closely matching of the grid interpolations (“Opr”) the step change becomes significant at the 10% two-tailed level in the mid 1990s and at the 5% two-tailed level in the early 2000s. Here though, the AWAP results are much more significant than both the observations and the existing operational analysis results, and significance at the 5% two-tailed level emerges by 1979. It is interesting to note that the drier climatology
Figure 6: Breakpoint analysis (one-tailed significance) on Perth RO (009034) annual rainfall. “PBR” denotes the parametric breakpoint method and “NPBR” the nonparametric breakpoint method as described in the text. “LR” denotes the (two-tailed) significance of the ordinary linear regression trend. “Stn” denotes the station rainfall totals over 1876-1991. “Opr”, “AWAP” and “HR” denote interpolated grid rainfalls over 1900-2008.
Bulletin of the Australian Meteorological and Oceanographic Society Vol. 23 page 32
has now persisted in Perth for sufficiently long that the linear trend (also shown in Figure 6) is statistically significant (something which cannot be said for Melbourne). 6. Gatton The University of Queensland site at Gatton (station number 040082) provides an interesting comparison from a different part of the country. Here, annual rainfall data are available from 1899 to 2008. Eight of the last nine years have been notably dry in this part of southeast Queensland, with the nine-year total being the lowest on record even though the driest of those nine years (2000) is only the fifth driest individual year in the record (Figure 7). Retrospective application of the breakpoint detection method gives 1941/1942 as the location for the most plausible positive break, from around 1945 onwards (until around 2000 where the two approaches switch to yielding earlier, but different, breaks), and climatologists might have diagnosed the increased rainfall as being statistically significant at the 10% two-tailed level from the early 1970s until the early 1990s. Several dry years in the 1990s cause the 1941/1942 positive break to lose significance. In stark contrast, the 2000s have been so consistently dry that the downturn in rainfall from 2000 onwards is statistically significant at the 10% level (two-tailed) from 2005 onwards and at the 5% level (two-tailed) by 2007. This achieving of statistical significance is even faster than that for Melbourne. It should be noted, however, that a reasonably wet 2008 has caused the significance to retreat somewhat to 14% (two-tailed) in the parametric calculation and 9% (two-tailed) level in the nonparametric one. Therefore, results obtained from the statistical methods used here are not in themselves sufficient to yield a robust
determination of permanent climate change (i.e., results that will never be overturned by the inclusion of new data). The double result in the Gatton time series suggests that interdecadal variability is important, and indeed this time series exhibits rather more auto-correlation than the Melbourne and Perth time series. Here the auto-correlation is significant out to a lag of four years (possibly reflecting the ENSO impact on the region), something which violates the assumptions of the break-pointtesting methodology (perhaps to the detriment of the significance estimation). The end of the year (December) comes in the middle of the wet season at Gatton. Looking instead at financial year (July-June) rainfall improves things a little – autocorrelation in the time series is reduced, although the lag four autocorrelation remains significant at the 5% level. The most plausible location for a single negative break occurs earlier in this time series, with 1990/91 (parametric) and 1992/93 (non-parametric) for the start of the new period and statistical significance at the 10% (twotailed) level emerging after around 10 years. 7. Concluding remarks It is anticipated that Australia will experience substantial changes in rainfall over the course of this century (Solomon et al. 2007) as a consequence of global warming. These changes in rainfall will mean that the regular cycling through droughts that Australia experiences will occur in the background of changing aridity. In recent years, a number of large and persistent rainfall anomalies have emerged (BoM 2008), supporting the earlier analysis of Smith (2004). Persistent below average rainfall has occurred across the southwest corner of Australia, across the southeast (Figure 4),
Figure 7: Annual rainfall (mm) for Gatton (040082), 1899 to 2008. Bulletin of the Australian Meteorological and Oceanographic Society Vol. 23 page 33
including much of the Murray-Darling Basin and in parts of southern and central Queensland. A natural question to ask is are these symptomatic of a shift to a new drier climate and how quickly might statistical methods forewarn of a shift in rainfall that achieves statistical significance. It has taken around ten years for statistical significance to emerge in the Melbourne annual rainfall time series regarding the recent dry period, when considered from the perspective of single break-point analysis, although the results from the grid-point interpolations yield slightly longer estimates for the time taken for this to happen. Even so, it appears that significance has emerged rather more quickly in the Melbourne situation than was the case with the now well-established decline in the Perth and southwest Australian rainfall. 8. Acknowledgements The authors wish to thank Drs Karl Braganza and Lixin Qi for their helpful comments in a preliminary version of this article. Appendix: Tests for normality, autocorrelation and data homogeneity Results for two tests for normality in the Melbourne Regional Office (RO) (086071) annual rainfall data (1856-2008) are given, since this is assumed in the parametric breakpoint assessment results presented in Section 4. The probability of obtaining a larger Kolmogorov-Smirnov (KS) statistic under a Monte Carlo simulation of 50,000 iterations was 0.77, while the probability of obtaining a smaller Shapiro-Wilk (SW) statistic was 0.73.
Figure A1: Quantile-quantile plot for Melbourne RO (086071) annual rainfall (1856-2008), as compared with the theoretical quantiles from the normal distribution. The line passes through the first and third quartiles (i.e., the 25th and 75th percentile points).
(The calculation of these test statistics was performed using the computer package R.) Neither of these probabilities is small enough to indicate a significant departure from normality. Figure A1 shows the quantilequantile (Q-Q) plot for the Melbourne RO (086071) annual rainfall. The data are approximately normally distributed, as evidenced by the Q-Q data being approximately linear. There is very little autocorrelation in the time series. The first significant (5% level) autocorrelation is at lag 9 ("9 = !0.200). The Melbourne observation site moved three times up until 1863. Since 1863 there has only been one move, from the Domain to the present location in 1908, although there was a notable change in the local site environment when a large building was constructed south of the site in 1996-97. A neighbouring rainfall site of good quality is Laverton Airport (approximately 19 km west from 086071), for which annual rainfall data are available between 1942 and 2008. This site is less affected by urban build-up than the Melbourne site. [There is no closer site available in the current high-quality annual rainfall network, for which details are available4, as is a site photograph for 0870315. Figure A2 shows the ratio of Laverton to Melbourne annual rainfall. Application of the non-parametric breakpoint test to the ratio time series gives 1956/1957 as the most plausible location of a positive break (statistically significant at the 2.6% two-tailed level), which
Figure A2: Ratio of Laverton (087031) annual rainfall to Melbourne (086071) annual rainfall (1942-2008), together with a quadratic trend line. The correlation between the two time series across this period is 0.85.
4
www.bom.gov.au/cgi-bin/climate/hqsites/ site_networks.cgi. 5
www.bom.gov.au/climate/change/map/stations/ 087031.shtml
Bulletin of the Australian Meteorological and Oceanographic Society Vol. 23 page 34
would be caused by Laverton becoming relatively wetter and/or Melbourne relatively drier. Restriction of the time series to the period 1960-2008 yields an extremely significant negative break (i.e., Laverton becoming relatively drier and/or Melbourne relatively wetter) at 1979/1980, although it could be argued that such a result is “cherrypicked”. The time series does not support the notion of an artificial drying at the Melbourne site over the past 10 to 20 years, and in particular shows no evidence of a breakpoint in 1996-97. A comparison of the Melbourne data with the gridded high-quality monthly data set (Lavery, et al. 1997) also shows no evidence of any inhomogeneity in 1908 or 1996-97. The high-quality monthly data set contains seven stations within 100 km of Melbourne, to the east, north-east and south-west. For the Perth RO (009034) annual rainfall (Figure A3), the probability of obtaining a larger KS statistic is 0.46, while the probability of obtaining a smaller SW statistic is 0.72. Again, neither of these probabilities is small enough to indicate statistically significant departure from normality. There were two significant site moves at the Perth RO (009034) site, in 1963 and 1967. In the case of Perth, the high-quality monthly set is unsuitable for comparison purposes, as it contains no stations on the coastal plain within 150 km of Perth, meaning that local values of the gridded high-quality set are likely to be influenced by inland stations which may not necessarily be strongly correlated with Perth. Instead, comparisons were carried out using a set of five stations within 50 km of Perth RO,
and with the single site of Perth Airport (009021). Both comparisons indicate that rainfall at Perth RO was anomalously low (by about 10%) in the five years leading up to the 1963 site move, but that there was no significant inhomogeneity between data after the two site moves and that at the old site prior to the late 1950s. References Bureau of Meteorology (2000). Climatic Atlas of Australia - Rainfall. Available from Bureau of Meteorology, Melbourne, Victoria. 25pp. Bureau of Meteorology (2008). Long-term rainfall deficiencies continue in southern Australia while wet conditions dominate the north. Special Climate Statement 16 (10/10/2008). Bureau of Meteorology, Melbourne. Available from www.bom.gov.au/ climate/current/special-statements.shtml. Clark A and Brinkley T (2001). Risk management for climate, agriculture and policy. Bureau of Rural Sciences, ACT. Fawcett R (2004). A long-term trend in Melbourne rainfall? Bulletin of the Australian Meteorological and Oceanographic Society, 17, 122-126. Fawcett R (2005). A reply to a comment on “a long-term trend in Melbourne Rainfall?” Bulletin of the Australian Meteorological and Oceanographic Society, 18, 62-65. Jones D A and Trewin B C (2000). On the relationships between the El Niño-Southern Oscillation and Australian land surface temperature. International Journal of Climatology, 20, 697-719. Jones D and Weymouth G (1997). An Australian monthly rainfall dataset. Bureau of Meteorology Technical Report No. 70. Bureau of Meteorology, Melbourne, 19pp. Jones D A, Wang W, Fawcett R and Grant I (2007). Climate Data for the Australian Water Availability Project. Australian Water Availability Project Milestone Report. Bureau of Meteorology, Melbourne. 37pp.
Figure A3: Quantile-quantile plot for Perth RO (009034) annual rainfall (1876-1991), as compared with the theoretical quantiles from the normal distribution. The line passes through the first and third quartiles (i.e., the 25th and 75th percentile points).
Lavery B, Joung G and Nicholls N (1997). An extended high-quality historical rainfall dataset for Australia. Australian Meteorological Magazine, 46, 27-38. McKeon G, Hall W, Henry B, Stone G and Watson I (eds.) (2004). Pasture degradation
Bulletin of the Australian Meteorological and Oceanographic Society Vol. 23 page 35
and recovery in Australia’s rangelands: Learning from History. Queensland Department of Natural Resources, Mines and Energy, Brisbane. 256pp. Nicholls N, Chambers L, Haylock M, Frederiksen C, Jones D and Drosdowsky W (1999). Climate variability and predictability for south-west Western Australia, in Towards understanding climate variability in south western Australia – Research reports on the first phase of the Indian Ocean Climate Initiative. IOCI, Perth. 237pp.
Pettitt A N (1979). A non-parametric approach to the change-point problem. Applied Statistics, 28, 126-135. Smith I (2004). An assessment of recent trends in Australian rainfall. Australian Meteorological Magazine, 53, 163-173.
Nicholls N, Drosdowsky W and Lavery B (1997). Australian rainfall variability and change. Weather, 52, 66-71.
Solomon S, Qin D, Manning M, Chen Z, Marquis M, Avery K B, Tignor M and Miller H L (eds.) (2007). Climate Change: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, New York. Available from ipccwg1.ucar.edu.
Nicholls N and Wong K K (1990). Dependence of rainfall variability on mean rainfall, latitude and the Southern Oscillation. Journal of Climate, 3, 163-170.
Timbal B and Jones D A (2008). Future projections of winter rainfall in South East Australia using a statistical downscaling technique. Climatic Change, 86, 165-187.
Bulletin of the Australian Meteorological and Oceanographic Society Vol. 23 page 36
Significant Mesoscale Oceanography News The recent grounding of the Shen Neng 1 on the Great Barrier Reef in April continues a recent trend of maritime accidents in the Australian region (Montara oil well in August 2009, Pacific Adventurer in March 2009 and Pasha Bulka, June 2007). Averting a disaster in this recent event was aided by the quiescent weather conditions and the rapid response of the Australian Maritime Safety Authority (AMSA). AMSA have recently conducted an inquiry to review the Pacific Adventurer incident (see link below). In the report, ocean forecasting services provided to the operators were identified as "highly accurate" although it was noted that the information was underutilised during the event. The recent SPILLCON 2010, held on 12-16 April in Melbourne, provided a forum to review the performance in recent events. There it was reported that ocean forecast services were actively used by aerial surveillance teams during the Montara oil spill. Montara Well Oil Spill - review The Montara oil spill that began on 21 August 2009 was finally brought under control on 3 November 2009 – a total duration of 74 days. On the day of the event, AMSA invoked the "national plan to combat pollution of the sea by oil", whereby the Australian Marine Oil Spill Centre (AMOSC) was activated. Within 24 hrs, the Atlas oil rig had caught fire and the crew were evacuated without loss of life. The first surveillance and dispersant operations took place on 23 August. Ship based containment and recovery operations commenced on 5 September and continued until 30 November. Australia's fixed wing aerial dispersant capability continued spraying operations from 2 September. Environmental shoreline assessments began in October. The oil well was finally capped on 3 December. The impact to the environment was fortunately relatively low with one affected sea snake and 29 oil-affected birds. There have been no reports of impacts to whales so far. Sheen was reported at Ashmore, Cartier and Hibernia reefs but no impact to shoreline or reefs has been observed. Environmental monitoring remains on-going. The oil spill continued over the 74 days at an estimated 400 barrels per day with recovery operations drawing an estimated 34 barrels per day and with a net leak of approximately 4000
tonnes per day. This makes the incident Australia's third largest oil spill disaster and less than 1/8 of the size of the Exxon Valdez spill in 1989. The environmental conditions during the event were relatively weak winds and clear skies. This period corresponds to the seasonal development of the warm pool in the Timor Sea, a pre-cursor to the Australian monsoon. Over the period of the event, the sea surface temperature (SST) of the Timor Sea steadily warmed by 4°C (Fig. 1).
Figure 1: SST for the Timor Sea on 29 September 2009 from BLUElink OceanMAPS operational analysis. These conditions permitted high quality ocean colour imagery during the early period of the incident that revealed the extent of early spill. However subsequent identification of oil slicks were hampered by small scale convection and natural slicks of spawn. The Asia-Pacific Applied Science Associates (APASA) provided AMSA with continuous trajectory modelling for the oil spill throughout the event. This information was routinely used by the operators to plan aerial operations. APASA had access to multiple estimates of current oceanic conditions from the Bureau of Meteorology’s operational BLUElink OceanMAPS, the US Navy GNCOM system and the CSIRO surface analysis system. The multiple sources provided guidance on the uncertainty of the forecasts, or so-called ‘consensus forecasting’. An objective study of the three products revealed that all were "best" for specific periods of event. The BLUElink OceanMAPS system was found to provide the "best" information for the greatest period (Dr. Brian King, personal communication). A review of the forecast trajectories for 30 August (shown in BAMOS, Oct 2009, Fig. 4) showed that the oil spill did change direction, with BLUElink OceanMAPS forecasting best.
Bulletin of the Australian Meteorological and Oceanographic Society Vol. 23 page 37
The close correspondence between the CSIRO analysis and the slick is due to it representing a hindcast. Overall, the ocean forecast information was found to perform well during the incident and was utilised throughout the operation. Eastern Australia The east coast of Australia has seen warmer than average conditions by up to 2-3°C. The central NSW coast in particular experienced warm and humid conditions in April. This is in part due to above average heat transport throughout the summer season but also the late seasonal weakening of the EAC. The current surface conditions for the Tasman Sea from the BLUElink OceanMAPS SST analysis on 26 April 2010 (Fig. 2), show 27°C water continuing to propagate down the NSW coast. The sea surface height anomalies (Fig. 3) show three large heat content anticyclones along the NSW continental shelf edge that are aiding the propagation of heat content along the coast. These eddies are expected to persist, indicating that warm conditions are also likely to continue for several more weeks. During early February, a coastal upwelling event took place off eastern Bass Strait leading to a large-scale fog event. This event was particularly noteworthy as it highlighted the difficulty of forecasting fog based on SST analysis products. Satellite remote sensing during this event provided poor coverage due to cloud in the region as well as the fog itself. The SST analyses were found to have no cooling at the coast. The BLUElink
Figure 3: OceanMAPS Sea surface height anomaly analyses for 26 April 2010. OceanMAPS forecasts, however, produced cool upwelling conditions as a dynamical response to the wind stress. The cooler coastal conditions were sufficient to produce fog, as observed. In the absence of satellite or in situ observations, it is difficult to validate the analysed SST's. However, the ocean forecast system appears to provide useful guidance for coastal fog forecasting. Western Australia The Timor Sea warm pool conditions have continued to persist into April, with large heat content over the North West shelf and anomalous SSTs. These conditions have supported warm, moist air over the Australian continent for an extended period. Further information: www.amsa.gov.au/Marine_Environment _Protection/National_plan/Incident_and _Exercise_Reports (Colour versions of these figures can be found in the online version of BAMOS – Ed.)
Figure 2: OceanMAPS SST analysis for 26 April 2010.
Bulletin of the Australian Meteorological and Oceanographic Society Vol. 23 page 38
Charts From The Past by Blair Trewin 13 January, 1984 Tropical lows (sometimes former cyclones, sometimes not) that drift across the continent are an occasional feature of the Australian monsoon period. These systems can bring very heavy rains to normally arid areas, with the mean annual rainfall sometimes being received in one or two days. An example of such a low affected Australia during the second week of January 1984. Forming as a broad trough over the Kimberley on the 6th, it remained near-stationary while intensifying, deepening to 995 hPa near Broome on the 9th without ever becoming a tropical cyclone. It then moved very slowly south-eastward while maintaining a central pressure in the 990-995 hPa range. By the 11th, it was centred north of Giles on the WA-NT border, moving to Kulgera in the far southern NT on the 13th, Lake Eyre on the 14th, and Broken Hill on the 15th, when it was captured by a front approaching from the southwest. It disappeared from analyses by the 16th. As might be expected, heavy rain was a feature of the low’s track. The most exceptional falls were in northern South Australia and areas of the NT near the southern border. Daily totals in excess of 100mm were recorded in this region on each day from the 11th to the 14th, with the highest daily total being 193mm at Muloorina on the 13th, the second-highest 24-hour January fall on record for SA. Granite Downs, north of Marla, received 357mm in the four days 11–14 January, and Oodnadatta 266mm in the six days 9–14 January, in a region which normally receives less than 200mm per year. Kulgera (150mm on the 12th) had its wettest day on record. Rain eased in this region after the 14th. The abundant moisture, however, interacted
with an approaching front to produce two areas of very heavy rain on the 16th. One was in northeast Victoria (193mm at Mount Buffalo, a January site record, and Baddaginnie), and one immediately north-west of Canberra (144mm at Fairlight and 122 mm at Uriarra, although only 53mm at Canberra). The front also produced high winds; a rescue operation was mounted on Port Phillip after competitors in a windsurfing race experienced difficulties. Severe flooding occurred in northern South Australia. The new Ghan railway, designed to avoid the flooding and washouts that plagued the old line, was closed for more than a week, as was the Stuart Highway. With earlier floods also closing roads in northwest Queensland, supplies ran short in Darwin, Alice Springs and Mount Isa, while Lake Eyre reached its highest levels of the post-1976 period. Major flooding hit Bogan in central NSW, while mostly minor flooding occurred further south. Southern NSW was to experience more significant flooding after a second major rain event at the end of the month, making January 1984 the state’s wettest month on record, as well as its coolest January. In South Australia it was the third-wettest month on record. Temperatures during the event were below normal but not record-breaking, though the air was extremely moist. The dewpoint at Oodnadatta reached 24°C on the 12th and 13th (although higher dewpoints occurred during the February 1997 floods). The rain and high humidity was damaging to those grain and fruit crops which had not yet been harvested, with 25% of the (bumper) 1983 wheat crop downgraded at a cost of several hundred million dollars.
Synoptic chart for 0000 UTC, 13 January 1984
Bulletin of the Australian Meteorological and Oceanographic Society Vol. 23 page 39
Calendar June
December
22–25 2010 Western Pacific Geophysics Meeting, Taipei, Taiwan.
13–17 AGU Fall meeting, San Francisco, USA.
28–2 July 13th Conference on Cloud Physics, Oregon, USA.
2011
July
11–13 18th AMOS National Conference (joint with NZMetSoc), Wellington, NZ.
12–16 11th International Meeting on Statistical Climatology, Edinburgh, Scotland. August 8–13 Meeting of the Americas, Foz do Iguaçu, Brazil. September 27–1 October Ninth Conference on Coastal Atmospheric and Oceanic Prediction and Processes, Annapolis, USA.
February
April 4–8 Greenhouse 2011: The Science of Climate Change, Cairns. June 27–8 July IUGG XXV General Assembly, Melbourne.
October 13–15 Australia-New Zealand Climate Forum, Hobart. 24–28 WRCP Open Science Conference, Denver, USA.
Australian Meteorological and Oceanographic Journal. Vol. 59: Special issue on ensemble prediction and data assimilation. March 2010. Articles: Abramowitz. Model independence in multimodel ensemble prediction Jones, Parslow and Murray. A Bayesian approach to state and parameter estimation in a Phytoplankton-Zooplankton model Fowler, Bannister and Eyre. Characterising the background errors for the boundary-layer capping inversion Wei, Toth and Zhu. Analysis differences and error variance estimates from multi-centre analysis data
Frederiksen, Frederiksen and Balgovind. ENSO variability and prediction in a coupled ocean-atmosphere model Frederiksen, Frederiksen and Osbrough. Seasonal ensemble prediction with a coupled ocean-atmosphere model Oke, Brassington, Griffin and Schiller. Ocean data assimilation: a case for ensemble optimal interpolation Andreu-Burillo, Brassington, Oke, and Beggs. Including a new data stream in the BLUElink Ocean Data Assimilation System Further Information: www.bom.gov.au/amoj
Bulletin of the Australian Meteorological and Oceanographic Society Vol. 23 page 40