 Thanks, Lee and thank you everybody. It's an absolute pleasure to be here and especially to be able to talk about the weather Which is one of my favorite things to talk about Not just the weather. I'm also going to talk to you a little bit about the future But before I do let's talk about the past. It's it's autumn now, right? It's the first first few days of March, which means summer is over and some of the data is in Here's a quick wrap-up of what we saw for summer behind you and Overall, Australia saw the sixth warmest summer on record Rainfall was not too far from from average. We were 2.6 percent below the average. That's averaged across Australia again I have the the top map is the mean temperatures So that's the average of the maximum and minimum temperatures averaged then across the summer to see How we ended up and what parts of the country were especially warm And I also have the rainfall deciles before that for the for the three months of summer So what's been driving our climate? Why why do we get the ups and downs that we get and why has our Winter spring and summer been the way that it has and probably the biggest reason that I'll flag with you is El Nino Which most of you've probably heard about by now Here I've got this the latest sea surface temperature anomalies Across the Pacific Ocean. This is for the week ending the 28th of February and that big area Does this have a laser pointer on it? Yes So this area across the Central Pacific that's very pink on this map This is areas where the ocean temperatures is warmer than it usually is for this time of year and this is This feature of warm water is what we call El Nino. We like to watch the oceans and ocean temperatures because they change Relatively slowly in comparison to other atmospheric features or atmospheric features that we have So it gives us a bit of predictability about what the upcoming months and seasons might look like by watching changes in the ocean We have This region in the Central Pacific we call it the Nino 3.4 box You know with a name like that that a scientist named it that Really doesn't mean anything except that it's this area in the middle of the Pacific and we average the sea surface temperatures there And we use that for an index of how strong El Nino is and I have the time series of how this has progressed up on the top left And the ocean temperatures are peaked in about November and have started a steady decline Back to more average or neutral for this time of year temperatures And so that's that's the progression of El Nino Everything in the shaded region is it is areas where the temperature of the ocean has exceeded a threshold that we've set to help us diagnose El Nino events The ocean temperatures aren't the only thing that we watch for when we're measuring this this slow change in our climate variability We also look for an atmospheric pattern and the the one that we look to most often is the southern oscillation index which is The essentially it's this the difference in the mean sea level pressure between Darwin and Tahiti which I've labeled as stars on this map and When we get a strong El Nino you get more low pressures happening over Tahiti and more high pressure over Darwin There's an old saying that what goes up must come come down And that's really what we're measuring the air over this central part of the Pacific is rising into the upper parts of the atmosphere It spreads out and it has to come down somewhere and it's coming down over northern Australia so we take the difference of the sea level pressures there and it gives us this southern oscillation index and I have this map or this time series of where that's been tracking recently and It's still even though it has a lot of wiggles in it. It goes ups and downs just like all the rest of us It's been in the El Nino territory for a while And I want you to remember this southern oscillation index and what it is and and how we can use it because I'm going to come back to this later on in my talk But before I do let's keep talking about the sea surface temperatures These are model forecasts for the ocean temperature in that Nino 3.4 box The white line is what we've observed All of these other very colorful lines are different models that run a similar projection forward But give us slightly different answers the dotted lines are our history so that's the the pink dotted line is the 97 98 El Nino the Yellow dotted line or the orange I suppose is the 82 83 El Nino and we can compare. How is this El Nino? Behaved in comparison with recent El Nino or recent comparable El Nino's and they all are giving us While not an exact answer some idea of what might happen in the future that as we look forward to about May or June We crossed this El Nino line into neutral territory And then beyond that we have three models in the La Nina space by July or August With all of the rest of the models in a neutral space So when you have the question of well, we're in El Nino now what comes next? This is the best guidance. We have to give you now that three out of the Seven models are saying a weak La Nina is possible. Whereas four of those seven are saying that That that neutral a neutral condition neither El Nino nor La Nina is possible We use the ocean temperatures to give us that bit that predictability because they do change slowly But we've got a bit of a problem that I'd like to raise with you and that's that the ocean temperatures have been changing so here here I have just a snapshot of January's ocean temperatures around Australia and These dark orange areas are places where the ocean is warmer than it's ever been on all of our records going back at least to the The late 70s when we've had good satellites to measure these things and using that same data We have a time series of how ocean temperatures have progressed not only using satellites But also using ship tracks ship measurements and boys across the Indian Ocean To give us an idea of how the ocean temperatures across the Australian region I don't label the region. This is just this Australian region on this time series How those have changed over the last century and the oceans are warming So here's the problem are we if we're using the oceans to try to predict the future But the oceans are now like we've never seen them before What does that tell us about our predictability of what's to come and how variable our climate is going to be? That brings me to the climate outlook and I'd like to take you on a bit of a journey about how we predicted climate variability Over the last several years and we'll start in June 1989 for those of you who remember June 1989 And a few big events that happened that year or that in that month was the Tiananmen Square Massacre in China Ghostbusters 2 was my new favorite movie and and the Bureau of Meteorology released our first Seasonal outlook and it was a little different than the seasonal outlooks. You might see today. Now. This is it It was issued in June for the months of June and July and it was it used the southern oscillation index as a predictor Which I've talked about already and it used just April and May and it found which parts of it with it The climatologist at the Bureau found which parts of Australia correlated well with rainfall and the SOI with a correlation coefficient of at least 0.4 or higher and With that correlation they were able to make this prediction of what the next two months We're going to look like and these were the only parts of Australia that we provided a forecast for because that was the only places that we had confidence in this outlook for and From there things improved quite rapidly. I'm not going to go through all of these But these are all of this step improvements that we had in our forecasting methods from 1989 until 1997 and What I will point out with all of these is that they all still used the southern oscillation index as the predictor of rainfall across Australia So it's basically using the El Nino pattern and the atmospheric response to that in October 1998 we took a Step forward in our technology of predicting rainfall We stopped looking at the atmospheric response and we stopped using the southern oscillation index alone and we started using Sea surface temperatures similar to what I showed at the beginning of my talk on the top I have a thing called SST 1 which is sea surface temperature 1 and on the bottom is sea surface temperature 2 SST 2 Remember that that'll come up again in a few slides But these are temperature patterns across the the oceans that we knew Or that we still know how Rainfall across the state Australia responds when the oceans look like this So the top one is a very well-defined El Nino pattern. The bottom one is warm Oceans off to our west coast and what we what we did is we had a statistical model that compared the The sea surface temperature patterns at the current or at the time with with these sea surface temperature patterns and using Mathematical methods empirical orthogonal functions, which will not be on the test. We were able to Basically compare how how do how do the current oceans compare with this and what does that mean for rainfall? And we were able to use that as a predictor This was also when we were able to issue our forecast two weeks before the next month Which was another step forward in our our predictability Going forward a little bit farther. We had we came to February 2000 which was the first time that we issued a forecast for minimum and maximum temperatures So I've got minimum temperatures on the left with all of these beautiful rainbow colors and maximum temperature on the right And this this was the first seasonal temperature forecast that we were able to give so at this point We're getting rainfall max and min temperatures in our forecast still using a statistical correlation with ocean temperatures to provide that outlook In September 2010 we took a little bit of a change again instead of using SST 1 which I showed earlier We began using the Nino 3.4 box as a predictor in its place that box I drew on the earlier slide We still use SST 2 in the Indian Ocean But we were able to show that using just that Nino 3.4 as a predictor gave us just a little bit better skill in our outlook So we started using that then and Then let's come to 2013 the dawn of a new era is what I labeled on the slide I'm probably making that a little bit more dramatic than it needs to be but it's a little bit fun in April 2013 was the last time we used a statistical model to forecast the climate variability for the next three months and Beginning in May 2013. We started using a dynamical model or a physics based model Which takes in initial takes in observations as an initial condition and runs the physics of the atmosphere ocean ice soil vegetation all of the physics that come into play in our climate runs that forwards for several months and Looks at how the how how will the climate be changing it just for the next three? What sort of variability can we see from that? and There was something else. I wanted to say about this, but a little bit a little while later the next year We used this new model. Oh, I remember what I was going to say was that this was the dynamical model is called Poemma, which is the predictive ocean and atmosphere model for Australia And it had been in development for almost a decade up to this point at which and it was when we were able to measure It had measurably better skilled in this statistical model is when we we moved over In August 2014 we updated our services in our website Of new a new and much improved website. It was interactive. It had more explanations. It had some monthly It provided monthly forecast in addition to the three monthly forecast It also had explanatory videos on it and it was mobile and tablet friendly and it was a It was a great step forward in addition to providing odds of being above or below the median rainfall or temperature It also provided a few other probabilistic Tools so the first one on the left is an outlook scenarios where you Can tell you you can say that I'm not really cared. I don't really care about the rainfall If there's only like a 25% chance that's gonna come let's look at the 75% chance mark And how much rain am I likely to get above that and it will give you that answer on the left I have a chance of at least where you can flip that and say I want to know How likely is it that I'm gonna get my 100 millimeters for the season or something like that and you can change that around They're all probabilistic based forecasts and for good reason too There's a quote by Wilkes who wrote a book about statistics and he said probabilities is the language of Uncertainty and none of us really know the future. Do we it's quite uncertain We take a model and we run it 33 times every week and we can overlay three weeks to get a 99 member ensemble Model outcome and they're all telling us something a little bit different and we can use that model output to give a probabilistic forecast But what are we supposed to do with a mop with a probability? What is a probabilistic forecast and how do you use it? Let's consider stacking a roulette wheel for a minute if we could If they're an equal number of red and black squares and you were to make a bet It wouldn't really matter if you picked red or black because there's an equal odds of landing on red or black But what if we were to stack the odds toward one or the other and we were to say there's more red than there are black You're not guaranteed to land on red But the more that you gamble the more likely you are to win if you keep voting or keep betting red every time So a probabilistic forecast is one That that predicts the odds of something that's going to happen here. I've got another example where we really stack the odds very high There's a 80% chance of red only 20% chance of black. We you can still land on black There's a 20% chance that could happen But I'd still I'd still recommend you you bet red with that because the odds would be in your favor and the more that you Bet the better off you'll be so what are you supposed to do with that? What do you do with a probabilistic forecast? Instead of red or black, let's switch that to above the median or below the median Here's an example from several years ago where the model was telling us that there was a 60% chance of having above median rainfall in southern WA, but what does that mean? It means that if you were if you were looking at a roulette wheel There would be 60% of the squares would be wetter than normal whereas 40% would be the drier normal than normal squares and So that's how you're supposed to use these probabilistic forecasts They're supposed to help you analyze the risk and understand how likely an event is going to be Although it doesn't guarantee that that event will come to fruition So with all of that in mind, I was going to show you the most recent climate outlooks. These were released only a month Oh, I'm sorry a week ago on the 20 27th of last month. No, that wouldn't have been right 25th These are the odds of exceeding our median maximum and minimum temperature with maximum on the sorry maximum on the left and minimum on the right and the the model is telling us that across the tropical north and Around the far south far south coast odds are favoring warmer than normal conditions for the next season for the autumn season This is looking at March April and May whereas This is that's for both maximum and minimum temperatures Whereas across the middle of Australia, there's nearly equal odds of exceeding or not exceeding the mean the medium This is now rainfall and we're able to break apart rainfall into the two next months And then the next season the three-month period odds are favoring drier than normal conditions across the tropical north in March and Odds are slightly in favor of drier than normal conditions across the south for the same month with near equal odds across the center when you look at April odds are Increased toward what are the normal conditions across the south sort of the eastern end of the Great Australian Bight I've had some discussion with people asking is that forecasting an autumn break and it's I'm not gonna say for sure because this is a probabilistic forecast remember But it is likely that we'll see that autumn break in March at least that's what the model is telling us Then you can look at the three months together together to get an idea of how likely you are to exceed your median rainfall for For the season below each map I have little green maps and these are these are hit rates. That's it basically a Measure of how consistent has the model been if we say that we have more than a 50% chance of exceeding the median and We really did exceed the median in that spot for that time then that would be a hit But if you didn't then that would be a miss We're able to run the model backward for the last 30 years to get an idea of how often we We hit or missed and we're able to show that statistic in these green maps where green is good and the white is not as good It's places where the model is essentially missed So what's next in the space of seasonal forecast? We want to understand the climate. What's what do we what do we see on the horizon? Well, there have been two major developments over the last year that Are interesting and will provide some great skill for us. The first was in the July to any 15 release of the Agricultural competitive white paper. There was Granted to the Bureau within that three point three million dollars for improved seasonal forecast You may have seen a media release from From minister Joyce about a month ago about this very thing so This is giving us the three point three million dollars to help us improve the seasonal forecast to give us better skill And a better service in addition to that we've received some government investment into a new super computer and This will be the new computer will become will come online later this year The the current system that we're using was last updated in 2013. It had 104 teraflops The new computer will begin with a 1600 teraflops and with a planned up upgrade in 2019 taking us to 5,000 teraflops And if you had to if you don't know what a teraflop is it's up here. I had to look it up It's big. It's how many How many calculations the computer can do every second in a teraflop is a trillion calculations, so This upgrade will give us five thousand trillion calculations per second. And what does that mean for? What does that mean for our seasonal outlooks? Well with with both the the infusion of funding and the better The improved super computer we're able to develop a new seasonal outlook model We're calling it access s which is an acronym for something the a is for Australia and we're life acronyms So the old model is poama 2 and the new model is going to incorporate some aspects of poama 2 as well as some aspects of the UK Met Office seasonal model and with both of these together We're going to be able to increase the model resolution from 250 kilometer grid to 60 kilometer grid and With with this higher resolution the the big step forward is we'll be able to better Resolve Australian topography, so things like the great dividing range and Mania Will now be will not be picked up in the model Just because we have a higher resolution. We're able to pick up that higher grid In addition to being able to resolve the topography better We also are able to pick up observations better. This is just an example of Australia of Across Australia the August mean rainfall in millimeters per day I have the observations in the middle with the old poama 2 model the one that we're currently using on the left and the new access s model on the right and you can see how well the new model is able to pick up the fine detail and on the Observations of rainfall that we have across Australia We estimate that this increase of resolution alone will increase the skill of our model by at least 10% In addition to the improved model will also be able to provide improved Services a few that that I'd like to point out Our goal with all of this is not just to have a shiny new model We want to provide you with the best information that we can possibly provide you and when you need that information So a few things that we're offering Pushing too many buttons is some tailored information for and briefing for your business and that includes training so if any of you are interested in Tailored briefing for your business. You can come and get a business card from me later And I'd be happy to talk to you about Taking some information that's just right for you. We can also provide a webinar service This is a screenshot of me providing a webinar If you have people spread out around the country and you'd like us to provide some weather and climate information to them directly We can provide that for you. We have informative videos and of course the new services will be mobile and tablet friendly We are promising to deliver a more accurate model and with the new model We'll be able to fill in a few of the gaps that we currently have so our weather forecasts end After seven days, but our climate forecast doesn't pick up until the next month And that gives you this gap between seven days and about four weeks that you really don't have any information to work with Well with the new model and the new improvements We're going to be able to fill in those gaps with some multi-week Forecasts and weekly updates of those to be able to fill in that gap during that very crucial decision-making time What else is possible? These are these are things that we could do but currently aren't Currently are not planned to do but there are things that we certainly have thought about in if resources allow We would like to go down this path and the the first one is providing other weather elements Rainfall and temperature are great But a lot of people want to know about things like humidity and evaporation Extremes when we're going to get a really hot spell come across the new model because it's not a statistical model It has all of this information in prison Intrinsic it's already built into the the physics of the model and they're there We just need to be able to pull them out and package them up in the right way This is my summary slide and it is identical to a handout that we have at the Bureau of Meteorology Booth if any of you want to take this summary slide back to your office with you with these improved Outlook detail on it come by the the booth and pick this up To summarize I'm gonna leave most of this for you to read, but I'm gonna tell you what my vision is Of our new of our new model My vision is that someday in the not too distant future probably beginning in early 2017 with rollout several months Following that is that a farmer will be able to go to one website and look at Forecasts exactly and exactly in line with the decisions that that farmer needs to make I'll be able to look at the day-to-day weather forecast to know if the rains are going to come How much and and where the winds will be and how much temperature will be? We'll be able to go and look at seasonal Sorry the multi-week forecast to be able to make strategic decisions about purchasing fertilizer and putting things down before a big deluge of rain or Protecting the crops against a massive heatwave. That's about a week and a half away We're able to look at the seasonal forecast and be able to make a strategic decision about what crops to plant over the next season because It will be good accurate information at the seasonal time scale And he'll be able to go to one spot and get all of that information and be able to make the decisions that he or she needs from that I'll leave it at that. Thank you very much