Short and long term forecasts

Growers can access seasonal climate risk information in 'The Break', newsletters and videos summarising rainfall, crop condition and potential yield for cropping regions across Victoria.

Making sense of seasonal forecasts and models

Farmers and our agriculture service industries and supply chains do an amazing job of making the most of our 'good rainfall' years to help grow record crops. However, increasing rainfall variability is putting greater pressure on farmers and supply chains to be more agile from year to year.

How often do we get average rainfall?

So do you know the key signals for when the season is about to turn? Do you know the key climate drivers behind the wetter and drier seasons for your region? When is the best time to look at forecasts?

Modern seasonal forecasting is a powerful tool for Australian agriculture, but used incorrectly or misinterpreted can also lead to poor decisions. These 3 videos feature some of Australia's leading seasonal forecast modelling and extension experts outlining the latest products, and how forecasts can and can't be used.

Billion dollar variability — The role for modern seasonal forecasting

Graeme Anderson — Climate Specialist, Agriculture Victoria — discusses how seasonal forecasting can help us better understand what we are in for, and plan for in the coming year. Graeme discusses the importance of climate literacy, to better understand what's coming out of seasonal forecasts and making sure we understand when and where these forecasts can be of value in our particular region.

Well thanks for that Dale. Graeme Anderson from Agriculture Victoria.

So our session today really is looking at this big topic of what’s the season up to.

And given the sort of billion dollar impacts from one year to the next of when we have big years or crook years, really exploring this role of seasonal forecasting.

The session we’ve got, we’ve got myself with the intro. Dale Grey who heads up the Break Newsletter in Victoria, he’s here. Dale’s an agronomist, so both of our talks will be coming from the agricultural side of things. And we’ve spent, you know, the best part of the last 10 years making sense of seasonal forecast and climate, and where this stuff can be used and where it can’t.

We then go into a 30 minute workshop where you’re going to be looking at seasonal forecast products that are currently there on, you know, international websites. And you’re in your table so that you can look at your part of the world. And we’re going to look at the May forecasts for this year. And you are going to see what they were saying in May. And then do a bit of a check to see, well this is what actually happened so how did they compare. And then it’ll context for looking at the forecast for the year.

One of the key bits is that reading these maps and forecast maps isn’t that straightforward. So on quite a lot of occasions people in agriculture are disappointed because they’re looking at the map and reading the wrong thing from it. So that’s the challenge today. The next bit, as always, if in doubt you should consult a doctor. So that’s why we’ve got Andrew Watkins here from the Bureau to bring home that final session and a panel discussion.

So seasonal variability has always been a big part of Australian farming, and this is Australian Wheat Production starting back at the sixties. And you can see, and you guys lived this, it’s always been variable. And as we saw last year, you know, when everything goes right, you know, we’ve got all of this; the best of our genetics, the best of our agronomy. When everything kicks in we can produce food than ever before. You know, and farmers and supply chains and advisors, everyone plays a key role in that.

But bad is still bad basically. So when you don’t add water all of that great stuff we’ve done is for nil. And it’s just interesting to me that when you look at that longer term, you know, there’s the potential difference between bad and things bottom out, and Australia as a country has had, you know, that it’s not raining anywhere as opposed to say last year when we’ve all had a cracker. There’s a bit of a gap there about, well what are you trying to plan for? Now this happens at a farm level but also at a supply chain level, about which are we planning for.

There’s a big difference there between our best — potential best and potential worst. And I guess, you know, talking to— in the Victorian situation we’d look at the Wimmera, 2015 was a shocker. You know, there wasn’t a lot of fertiliser needed that year. And I can imagine the discussions that happened at the end of the year along the supply chain with people caught with too much stock. And the bean counter saying listen, don’t do that again. Now the very next year the Indian Ocean Dipole was in its wet phase. And you know, this is a one in 20 year event to capitalise on. And so that relies on, not only the farmers having everything ready to go, but also the supply chain there to support them.

And it was interesting that we did a Twitter survey in 2015 and two out of three farmers said they had trouble getting inputs, whether it was fertiliser or fungicides to deal with this massive season. So it puts a lot of pressure, not just on the farmer, but right along all of the brave people who are making calls about, well what are we planning for this year— from one year to the next.

And is there a chance that seasonal forecasting can help give us a bit of a heads up. So that’s what we’re going to try and share with you today. And when we talk about forecasting, and we do, you know, about a hundred sessions a year Dale and I with farmers around the place.There’s three things that normally pop up.

People say we want better forecasts. Well, do you think the forecasts are getting better? Yes. Yeah. So, you know, what we’ve got on our phones, and when you look at, whether it’s seven day forecasts or whatever, they do get better. I’m really glad we’ve got the Western Australians at the front here, because — and it’s not so we keep an eye on you, because they’re going to support us and ask good questions, so thank you. So forecasts are getting better.

And there’s bigger computers, more data getting fed into them, so that’s all good.

But that’s not all. One of the key things is better climate literacy, so we actually understand what’s coming out of forecasts. That’s a really key one, because there’s a lot of people disappointment with forecast is often to do with misinterpretation.

The other bit is it doesn’t matter what the forecast does, it’s well what decisions people make with it. Now that’s what you’re good at, that’s what farmers are good at. So really this session today is really looking at this literacy thing just to try and make sure we understand when and where these forecasts can be of a bit of value.

So we need all of those things happening. So, a bit of a quick recipe. So if you’re on Master Chef and the special ingredient tonight is we’re going to make rain, there’s only — there’s two really key things we need to make rain.

So does anyone know what they are? Two really key things to make rain. Without them you haven’t got a chance. That would be water. Water, that’s right. Well, you’re quite right. Water. But before you get to the water bit, you’re looking up in the air, what do you need in the air? Clouds. Yeah.

So what’s in the clouds? What’s the clouds made up of?

Moisture. Yep, moisture. So you’ve got to have moisture, all right. So if you’re starting off with dry air your chance of making rain’s pretty slim, right. So you’ve got to have a source of moisture.

Now this is just where it’s rained in the last six months. But I remember a meteorologist telling me if all we had was moist air it would just be humid. So something else has got to happen to make that moisture fall out. What’s got to happen?

'Condensing of the droplet.'

'Yep. What does that?'

'Dust or something.'

'No. Temperature. Yep. Right.

So it’s really interesting. You look at here about where it rains and you’ve got the Great Dividing Range, you’ve got moisture coming in off the ocean. That moist air gets lifted up, as you get up against colder, and basically that colder temperature makes rain fall out. And it’s interesting, when you go on the other side of a mountain range, when air is coming down and sinking, as it’s coming down it gets warmer, can hold more moisture. And that’s the sort of the rain shadow effect.

So really, the rain that falls, you know, over our heads is to do with some moisture’s met up with some cooler air and it’s falling down. So that’s the key bit.

And the seasonal forecasting is largely around, is this one of those seasons where there’s more moisture about to feed into the rain making process for our part of the world or not. And then where are those sort of triggers for rain.

So I really like this picture because it’s, sort of, got, you know, a lot is that Antarctica down south. You can see all the cold fronts, the conveyor belt that wizzes around the Southern Ocean that’s a Southern Annual Mode.

The big sources of moisture though coming out of the warm oceans up north, so the El Nino southern oscillations, so the Pacific Ocean is a really big source of moisture for Australia, some years it’s helpful, some years it’s in a bad mood.

And over here the Indian Ocean’s dipole, and so some years it’s really helpful like it was last year and other years it’s not helpful at all, so those two up the top there are really key drivers. And they’re not active every year but it might be every second, third or fourth year. If they’re doing something it’s worth knowing about because saying the supply of moisture is either about to be enhanced or it’s about to be, sort of, cut off.

And down south for us it’s just really about where those fronts are all sitting. So a lot of what sits behind forecasts is about, well what sort of form are these climate drivers in.

And we’ve sort of had analogies and when Dale — when we put together the break a bit like the mobile phone and the reception when you’re driving around the countryside. Sometimes you’ve got a five bar reception and sometimes it slips out to nothing. Well if you talk to the seasonal forecast meteorologist, we’ve got some years where there’s a five bar forecast and some years where it’s down to nothing.

And some years where they’re quite confident about that five bar forecast being around for a bit longer. So part of that is trying to understand, well what are those signals for our part of Australia; which are the bits of ocean we should be looking at.

So where do we find the most moisture in the cloud?

So if we’re looking across the planet where will that be?

Equator. Fantastic. Thank you Nick.

So a lot of the meteorology, and especially with seasonal weather patterns, is all about looking at what’s happening along the equator because it’s the hottest part of the planet. And if there’s going to be anything a bit unusual or different happening, it’s going to be because at the equator some warm water’s popped up somewhere and that is where you find all of the cloud and moisture is above that.

Now if that happens over our part of the world and all the moisture is heading our way, if it’s happening somewhere else then that can drag it — drag these weather patterns out of whack. So that’s what is sort of sitting behind a lot of these forecasts.

And here you’ve got the classic sort of El Nino cycle where, you know, the warmest waters out in the east and a lot more cloud activity shifted out there. It just makes it harder to — you know, there’s less moisture coming in on those years. And then some years it’s all happening to our north. So that’s what is sitting behind seasonal forecasts.And one of the things that, when they’re looking at these sea surface temperature maps.

And this is a sea surface temperature anomaly map. Along the equator is actually a lot warmer water. It’s like a warm bath. And down here it’s absolutely freezing. But this is an anomaly. So it’s showing, you know, these colours are where the water might be two degrees warmer than it normally is, and the blue colour is where a bit of ocean might be two degrees cooler. So this is a drier pattern for certainly eastern Australia.

September ’06, it’s one of those bomb out years from a wheat point of view. El Nino over in the Pacific, warmest waters popped up there. There’s extra cloud and everything happening. But you’ll notice over to the north of Australia those cooler SSTs there, so more moisture’s happening out here, less happening here. And there be more clear skies and everything, and the trade winds would be heading the wrong way, and all of that is so not helpful for us.

Now if you look at this sort of area just at the north of Australia and we’ll say spring 2010 for eastern Australia just see if you notice any difference.

So there’s the difference. So a lot of the season forecasts are saying well where is this warmest water going to pop up? So back there, you know, there was, in that spring, plenty of moisture up there. So a lot of the season forecast is around is there anything major going on with these climate drivers? So it sets up a whole chain of things. And if we look at last year’s SST from July, again you can see it was really quite warm up here. But, you know, all of that, the moisture sources were set to four or five, pretty good. And also for southwest WA you had a bit of cooler water there which is part of that local pattern which is really useful for you.

So it’s interesting, if you look at the bomb out years on that wheat production map, and I’ve just printed off the spring SST anomalies, you’ll notice what’s common with each of those. You can see ’94 was looking a bit blue — 2002 was looking a bit blue there — and a bit blue there in ’06. So part of that seasonal forecast has been there’s some signals there about when things are perhaps going to be less favourable and that’s what feeds into these seasonal forecasts.

I had a farmer say once about seasonal forecasting saying, you know, it’s really a bit like getting the minutes of the latest Reserve Bank meeting. It’s, you know, it’s not so much the prediction but it’s more the commentary which is really readable. And quite often they haven’t got much to say. But every now and then there is something there where, you know, when you look at their previous forecasting and then what’s happened this time, you say no, it’s actually starting to happen. You can have more confidence that the economy is going this way or that way.

I like the — I’m not sure if anyone’s seen it, the Checkout Show on the ABC. And they’ve got this Product versus Packshot, where you know, this is what’s on the product and open it up and that’s what’s inside the package. I always get a bit of a giggle out of that.

But it’s interesting from a seasonal forecast point of view. I think it’s really important our own expectations because we’re often looking at this. Looking at a forecast map and thinking you beauty, and then three months later we’ve had to deal with that thinking there’s a bit of a difference there. So that’s why a really big cautionary tale about getting our expectations right of what a forecast can and can’t do.

A nutritionist would say that, yeah, well the corn is there. The corn is there, lentils are there, the sour cream is there. It’s officially right. But our expectations really extract two different things from those two slides. So our own expectations is important.

And I’ve got a couple of one-liners here which we’ve collected over the last 10 years I guess, which I think are useful about how people use them.

And you know: 'Make decisions on what’s knowable first' which you know, most of our agronomists and everyone does. You know, soil, moisture, what’s in the  bank, all the other stuff, that’s the place to start. Seasonal forecasts are a value add, okay. 'All options are possible at any time, but some seasons a pattern can change the odds of what’s more likely.'

'A 70% chance of wetter than normal is the same as 30% chance of drier.'

So there’s a lot of people did — if you see a 70% chance of wetter then lock it in as going to be wet. But actually 30% chance of something going the other way is still significant enough. So it’s important to know what the map is you’re looking at and what it’s actually saying. 'Best to squint your eyes when you’re looking at a forecast map, it’s the vibe you’re after.'

Now as some of these maps get better and more accurate, I think if you’re looking at it in your district saying, oh the Wimmera’s going to be wet but the Mallee isn’t, you’re looking too close. You need to hold these back. Because it’s really about these are big, big climate drivers, and it’s about your sort of part of the world. Is it looking at a drier pattern or wetter? Individual weather events will decide whether the Mallee gets something or the Wimmera gets it. But this seasonal forecasting is more about that bigger scale. And we’ll be having a crack at that in a minute.

'The commentary that sits behind the forecast is a really a key bit.' And that’s why a lot of people look at the map but just spend a bit of time for that extra paragraph or two about what it’s saying. I’ve heard this one. 'Just one weather event can bugger up a good seasonal forecast.' And that’s what we always hope for when there’s a dry forecast and you can crack one decent event that has made the season.

And that’s what farming’s all about, you’re backing your judgement. But seasonal forecasts, you know, that’s the pre — the set up conditions for the year. Individual weather events determine whether that’ll happen or not. A key one on literacy. 'Knowing which climate drivers affect your regions, big droughts and wet periods.' That’s been better known than ever before.

As to where climate change fits in, climate change will turn up one week, month, season and a year at a time. So seasonal forecasting is still only going to become more important, especially when, you know, a bit of warmer water than we’ve seen before pops up at a key part of your place, you can see sort of more moisture coming in.

And, all models have got value and so we think it’s useful to look at a bunch of different models rather than just one opinion. So part of what the workshop is will be doing that.

So thanks. I’m not sure if one of the sponsors that’s sort of helped with some of the extension on this is the Managing Climate Variability program, so I’m just going to give it a plug. It’s got a range of industry partners in there. Works pretty closely with the BOM and climate researchers on making sure that this stuff does get in the hands of people who are making the big decisions like you guys.

And if you don’t get the Victorian, The Break, that does every month there’s the email, it’ll pop out free.

And also with what Dale’s going to share is the forecast product he looks at when he puts together The Break. And while we do that, the Victorian it’s got a pretty big following, we’ll be quite happy to help work with anyone who wants to — say do that for WA or Queensland or what else, we’re happy to train you up, so that’s part of the course.

Insights for interpreting seasonal forecasts

For the past 10 years, Dale Grey — Seasonal Risk Agronomist with Agriculture Victoria, has been putting out a summation of what the climate models are saying for Victoria. In this video, Dale outlines his experience with interpreting seasonal forecasts.

So Graeme and I, we’ve been working in this space for probably 10 or 11 years now, and for the last — all of those 10 years, at every month I’ve been putting out a summation of what the models are saying for Victoria.

So I’ve been looking at models and their outputs for some time. First of all, just a little sort of assessment explanation of what a computer climate model is. It’s a mathematical representation of what the physics says is going on around the world. But it’s of a really coarse resolution.

So here’s the ocean surface in here. You’ve got a square in there. There’s mathematical numbers going on to describe that ocean for that whole box of about a 250 kilometre square. And above that square of ocean you’ve then got a heap of stuff going underneath that, so lots of layers of boxes with mathematics describing them underneath. And then above the land you’ve got — you’ll have the land surface with mountains and stuff like that in it. And then above that you’ve got lots of layers of the atmosphere showing how high that’s going up. And all those things are, you know, working with the laws of physics; hot air rises, warm air holds more moisture.

There’s salinity, there’s currents, there’s winds, all those things going on. But the resolution is pretty coarse. Now what’s interesting is that our computer power — computing power keeps getting better and so resolution keeps getting smaller. And if anything though, our ability to initialise these models is better than we’ve got of mathematically representing them in the model itself.

So once every two or three days we have data coming in from all these floating Argo buoys. You know, once a week they come to the surface and tell us what’s going on underneath them.

Satellites zipping over it daily giving us information as well. Permanent buoys situated all through the Pacific going up and down every day to the bottom and telling us what’s going on in their little patch of the world, with a wind finder and temperature depth. And then, yeah, the people that are taking your reverse ships from WA and loading wheat back to Saudi Arabia or something, give them a few bucks on that ship to take the temperature to do that. And they’ll do that because they’re bored out of their mind. And we’ve got fantastic information along the shipping lines of the world about temperatures.

So you add all this stuff together, that then has the ability to initialise the model. But the resolution of this information is much better than the model’s actually capable of currently doing. So let’s have a look at some models here.

This is the Bureau’s POAMA model. That’s the current model. On the other side we’ve got what we’re moving to in the next, or this year we would hope. And this is the UK Met Office’s Model. The current Bureau model POAMA works on a 250 kilometre square grid, so really coarse resolution. And the new big super computer is now going to be capable of working at a 60 kilometre resolution. So what I propose to you there is that on the current model here things are saying they’re a bit wetter here around Sydney. But if you sort of said, oh how about up around Ballina or something? Can you make a big difference between what’s happening here to up here? There’s a slightly — I think there was supposed to be a slightly wetter section here around Albany.

But could you actually say that that’s going to be wetter than Esperance out here? So this is the question about we’re saying. Is to take a step away back from the model, cross your eyes, wait for the sort of 3D puzzle to happen and go with the vibe of what it’s saying. Because the resolution of this thing is so coarse that you can’t make really close distinctions between properties. However, when we get to a 60k resolution well we have a much better ability to do that.

In fact on a 250k resolution you can’t see the Great Dividing Range. There’s not enough actual resolution in that model to see that. On a 60k one you can. And hopefully the picture is good enough.

You can see a rain shadow effect down the Great Divide there from this prediction.The other thing that’s most certain, or not certain, is it’s critical to know the climate drivers in your area and how they affect the rainfall and at what time of the year. Because that varies around Australia and models are much more believable when they’re producing — they’re putting out forecasts. And they’re forecasting things that are capable of happening at the time that affects your rainfall. It just gives you a lot more confidence.

So here’s something, this is for central Victoria here, around Bendigo. The ticks in the boxes show when these particular climate drivers have the greatest correlation with rainfall. And fortunately that’s through most of our growing season, from May through to October. We’ve got Southern Annual Mode doing a little funny flip thing in summer.

But what people and farmers are always saying, well what the heck’s going on there in summer.

And particularly in autumn, and it’s what we call the dreaded autumn predictability barrier, where predictability at this time is never good. And it doesn’t matter what the model is in the world. It’s never as good here as it is over here.

So take least notice of models when they’re making predictions at a time where those particular climate drivers don’t affect your rainfall. Or if they do it’s very hard to tell it from a year where they’re doing nothing. So is your model reputable? Now this is — I’ve never put this slide up before, but I reckon it’s pretty cool. Because is the person putting out the model information? Do they come from someone who’s actually capable of doing this sort of stuff? This is 50 petaflops a second or something. This is the old 486. The person using this has no hope of doing the stuff like this on this resolution. And only a massive computer like this is capable of doing it at 60k resolution. Which you might say is not good enough either, but that’s just our lot in life, and that will slowly get better.

So does the computer publish skill information. Or the model agency, do they produce information where they’ve tested their model over time and they can tell you we know where it’s good and we know where it’s poor, and it’s better at these times of the year than others, as opposed to not telling you that information at all.

Does it provide hindcast? Does it give you information to show how that model’s worked in the past? And you can go and have a look at wet 2010 and going did this model even predict that? Oh look it did. Or gee, it didn’t. Gee it’s a bit strange everything else was. And really critical is the best thing about some models is they provide some commentary about the model says this.

We think that’s really strange. We think this is going on x, y, z. As opposed to there’s your map. Suck it and see it and stiff bickies. So this gets us to the concept of skill. And what I’m going to do now is present a heap of information from a number of different models, how they present skill of their model.

So these are some skill models from the Bureau of Meteorology’s POAMA model just for Australia. I don’t think we’ve got anyone —one person sitting from the overseas section over there, so we’ve left them out of this unfortunately. But what you’ve got there is the skill of POAMA’s model at varying times of the year, in summer, autumn, winter and spring. And the skill ranking from low, medium’s not bad, high is pretty good. And what you can see is if you’ve got any kind of eyesight at all, and even Graeme with red/green colour blind can pick this up. But unless you live in Rockhampton, nowhere in Australia has high skill at any time of the year for this model.

And you can probably pick some areas like — not that anyone’s really farming above the Great Australian Bight, but you can see the model never has any skill in some of those areas there. So when you look at a forecast it’s critical to look at this information to say should I be making — taking more notice of what’s going on here. Or is this a time where, you know, they’re not usually very good, and it’s interesting what they’re saying, but I’ll just kind of put that to the side a bit.

Here’s a different way of presenting skill. This is from the ECMWF in the UK. And the first thing for the South Australians and — sorry, the Victorians and the Taswegians and the New Zealanders in the room, is they’re going what’s happened? Where the rest of the world?

And so what the model’s interesting — or does, is it believes it doesn’t have very good skill for predicting in these lower latitudes so it chooses not to show you it. You can’t even see the prediction. If you’re from the Bureau of Meteorology you can but you’re not allowed to tell anybody else. The other thing they have in this thing is this kind of statistical test. I’ve got no idea what this is and so that’s always hard to make comparisons when they don’t tell you how they’re actually doing their stats. But when an area is shaded in they’re saying there’s a 10% something that’s good, and when you get a line that’s bordering the shading that’s telling that there’s a 1% level of some sort of skill.

So when you’re inside the boundary borders that’s clearly better than just being shaded.But that’s the way this model chooses to represent whether there’s any skill in the prediction that they’re making.

Here’s another one. Here’s the — this is the UK Met Office, and they use a thing called a ROC score. And basically anything above a half is good; anything below a half is useless. So the grey colours are useless, the orange and red colours are really good. So when you’re assessing the skill on this model more orange for that time of the year the more skill.

Here’s a model from the US. And it uses a different thing. It uses anomaly correlation, which Andrew was telling me is a pretty poor way of doing skills, so take that in mind. But a correlation, anything above 60% is not bad. So looking at this model’s output you can see some pink and dark green in a few areas, but you know, the skill there’s not exciting over much of the world which brings us to now looking at models and how they represent rainfall.

There’s two ways models do this - probabilistically and deterministically. The probabilistic ones are the ones we’re used to seeing from the Bureau of Meteorology. And we see this one, the old chance of the rainfall being above median, a probability of above median. And as Graeme was showing before, when they’re saying there’s a 30 to 40% chance of it being above the median, that’s really saying that there’s a 70% chance that the rainfall is going to be less than the median. The median being closer to the average.And so what that is saying is there’s a greater probability of not of it being drier.

But what that output there does not say it will be dry. It’s not an emphatic thing like that. It’s couching things in terms of probability.

The next way you do it, same data, same time, this is called a Tercile distribution. Tercile means three. And, you’ve divided the data up into dry, an average third and a wet third, and you’re getting a probability of where it’s going to fall in between that. So giving you more information than just above and below the median, but still, once again, couching things in terms of probability. And not giving you anything absolutely definitive.

Here’s another way that the ECMWF does probability. And they’ve got their — you can see there a 50-60% chance. What is — yeah, so they’re just — it’s more similar to those other ones. Yeah. This is a tercile one as well. So you’ve got a chance of it being drier.

If it’s white you’re sitting on the fence. And if it’s green it’s predicted to be wetter. But there’s also the opposing probability of being dry as well.

Then we get to the deterministic ones. And everyone loves these because these will nail their flag to the mast and go I think it’s going to be dry. But it’s very easy to take these wrongly as well, once because of that resolution issue.

The other thing you’ve got here is that the scales of these things vary. So this is a model for three months. A modelled output for three months for winter. And the scale here you can see is in 0.6 of a millimetre per day; negative 1.5 millimetres per day. You’ve got to multiply that by 90 in your head to work out how many millimetres it is. So this is sort of saying 27 to 54 mls less than the average, so a drier forecast.

But the critical thing is not to get hung up on how many millimetres is it saying. Is it the difference between me putting a crop in or not? You know, it’s not good enough to do that. This is another way. This is the ECMWF’s way of doing it. And so there’s 100 to 50 millimetres. They’ve actually given you the aggregate.

You’d have to multiply it by 90. They’re just saying that it’s 100 to 50 millimetres less than normal when you’re that colour. So that’s relatively easy to look at. But as I said there, don’t get too hung up on the numbers. And go for the vibe about whether things are looking wetter or they’re looking drier.

Righteo. So here’s a map — this is my fast break analysis of lots of different models.

For spring 2015 and done in autumn— sorry in August. And you can see across the board there every models predicting an El Nino. So an incredible consensus of the models that they reckon that was going to happen. But in terms of rainfall, you know, half the models have got drier tendencies there. And the rest of them were sitting on average.

And then, as we know, what turned out was it was particularly dry over the southern parts. Match that up with 2016, a completely different season. A chance of a La Nina perhaps, but every model’s saying negative IOD. Half the models once again going wetter for spring, half on average, and as we know it turned out to be quite wet. So consensus we think is much better than just going and picking one model and hoping it’s right.

The models have skill but you need to know what times of the year to be looking at them. It’s bit like Kenny Rogers, know when to hold them and when to fold them.

And they had poor skill at some times and not in others and location does matter. So I’ve got there — I’ve got my picture of Dennis Denuto. I had to learn how to spell Denuto right to get it up, but it is the vibe. It is about what is a number of things saying here rather than just picking my favourite model and hoping it gets it right this year. And as Graeme said, if you’d like to do this for your state or region, we’d be happy to train you and help you out. Thank you.

Latest on seasonal forecast products and outlook

Dr Andrew Watkins — Climate Prediction Services Manager, Bureau of Meteorology, describes the science and complexity around climate models, and the incredible amount of super computing and interpretation that goes into the BoM's forecasting. At the end of the day, it's all about probability.

Thank you. Thank you Graeme. What is it with us and showing pictures of snow? It’s been a great snow year if you’re a Victorian or NSW person.

So what’s climate and weather? What’s the difference between climate and weather?

So in other words what’s the difference between a climate outlook and a weather forecast?

Well basic climate is about long-term averages and slight pushes or shoves one way or another from those climate drivers. So I sometimes say look, a good example is climate tells you what clothes to buy. So if it’s winter I’ll buy some jeans and few jumpers. But weather tells you what clothes to wear.

So if you’re up in Sydney yesterday you certainly would have been in tank top, the Reg Grundys and shorts a pair of thongs because it was about 34 degrees in Sydney yesterday. So the other example there, so for Geelong September the 14th in Geelong averages 16.5-17 degrees. And today was forecast to be 15.

So the climate would tell you one thing. I should say that the calendar, your calendar on the wall is a climate model because it tells you effectively what the average climate will be. You know it’s going to be cooler in June than it will be most of the time in January. So, you know, climate models are lots of different things. We use something slightly better than a calendar.

And, so hopefully this runs. So in terms of the climate models that we use to do the climate forecasting, we’ve heard about POAMA, but it’s really just — I think it’s predictive ocean atmosphere model for Australia. And, these are big, as Dale showed, big dynamical, big things based on physics and mathematics. But the reality is, what they do is they generate weather. So this is from the climate model, going through every day. And it does a forecast every day out for, in this case nine months ahead.

But not just weather in the atmosphere, but also in the ocean as well. It might be subtle but you can see eddies and currents and so forth in the oceans as well. So basically we run this model, this physics of the atmosphere and the oceans, and also of the ice and even land processes like soil moisture and so on. Some models even have chemistry within them, so you can look at how different gases in the atmosphere may change over time, say as — like in the northern hemisphere where the trees lose their leaves or gain the leaves, that changes the amount of carbon dioxide in the atmosphere and you can actually see it wobbling up and down.

So these models are quite complex. They run on the big super computers. They cost many millions of dollars to develop. And the super computers themselves cost tens of millions of dollars. And then they cost lots and lots of power. So in other words, basically at the Bureau the humans are irrelevant in terms of how much electricity we use. Most of the electricity is actually used by our super computer, both running it and also keeping it cool.

So we run these models out. But we don’t just run them once because strange as it may seem, there’s not just one possible future in terms of weather. Because little random things can change what may happen in the future. We actually go and run the model 165 times So we do a forecast every day for nine months ahead and then we repeat it 165 times just changing —us changing the starting values ever so slightly.

So let’s pretend that say the thermometer was 16 degrees, well we’ll run it at 16, we’ll run itat 16.1, we’ll run it a 15.9. We’ll change these variables slightly to try and pick up the potential weather states in the future. And then the proportion of these 165 that give us wet weather is effectively the seasonal outlook. So if 70% of our model run said it was going to be wet; 30% said it was going to be dry; our probability for the season ahead would be 70%. So, it’s quite different to a weather forecast model which is really just giving you these daily patterns and are usually only run once.

So you’ve got your probability, so what does that actually mean? So if this is our seasonal forecast and the chance effectively, well this is roulette wheel really. You know, each year we’ve got a roulette wheel that’s been changed. So it’s got around about the 50-50 chance; 50% of the slots are red; 50% are black. Well we can see in that one it fell on the red. If we hopefully spin it again with the odds 60 and 40, well we ran it that time and okay it came out as — it came out as red.

So you’re not guaranteed of the last one — you’re not guaranteed of the outcome but you’re favouring it. So hopefully this time we’ve got 80% red, oh, it fell on the black. So it’s all about probability. Now what we’re really doing, or what I think Graeme said before about the Reserve Bank and so on, we’re effectively giving you those odds that just push things a little bit in your favour.

Now the last time I looked at various billionaires and so on that own casinos, and we won’t mention because we’re being filmed, but they’re billionaires right. They own casinos. They only have such a slight odds in their favour. They only have a couple of percent of the odds are in their favour and yet they’re the ones driving the Rolls Royces— the punters are going home on a skateboard. So in other words, you only have to have things that are slightly in your favour to give you a benefit in the long term. And that’s what we’re suggesting here, that if the odds were in your favour in the long term you’ll come out ahead. But on the year to year or day to day scales you can get the opposite of what you might expect. It’s all about probability.

So here is basically our seasonal outlook for spring. We’re looking at point here in southwest WA. So there’s your bar at 60% chance being drier, but there’s still a 40% chance of being wetter. So you can think of that as your roulette wheel. It looks a bit like this one. Spin it round and round it goes and where it lands, well hopefully it will be 60% but really nobody knows.

So in terms of the climate outlook and accuracy, and this is where we’re talking a little bit about the difference between accuracy, your skill, and the confidence we have in the outlook. So on our — with our Bureau maps, or with our Bureau outlooks, we always have these plots of skill. These maps are basically just saying historically, if we look at the past how well did it do?

And we can see that where it’s greener we’ve actually had more skills. So the greener the blob the better we’ve done.

But the reality is, in terms of confidence, and we know the skill where it does better historically. So if you ran the model back in 1980. You ran it in 1990 and in average how did it go, what we actually see is that when we have strong climate drivers such as El Nino or La Nina, so those strong, those big patterns in the Pacific Ocean that really drive our climate we actually do better. So we actually have higher skill in those years when we have the big climate drivers.

And we can see, look it’s only a subtle difference when we have El Nino and La Nina, we do overall on average a bit better for Australia. But certainly many of the individual events that have the highest values are when we have the strongest climate drivers.

When we have an El Nino we tend to do very well. When we have a La Nina we tend to do quite well. And when we don’t have much of a signal at all we tend to have more — things tend to come back down to the pack and we don’t do as well, although we still do better than guessing.

And so I guess last year — last year was a good example in Victoria at the very least.

I had the joy of going up to Birchip a few times during the year and talking with the property group up there. But in April when I visited and they were dry sowing and so on, so Birchip here in the Wimmera Mallee, it was pretty hard to walk around literally with your boots bringing up the dust, trying to tell them that the seasonal outlook was actually generally favouring wetter than normal conditions.

And we were able to say to them, look, you know, the IOD, the pattern in the Indian Ocean that favours, or doesn’t favour rainfall for Australia, but the Indian Ocean patterns are really favouring wet conditions for Australia. And ultimately, even though, you know, you’re kicking up dust, you’re finding it a bit hard to tell them it’s going to actually — I reckon it’s going to get wetter, it did of course become wetter during the winter months through large parts of Australia, largely driven by the Indian Ocean pattern and a very strong signal that that was giving.

There was a lesser signal from the Pacific Ocean although it was at the end of an El Nino and typically at the end of an El Nino you get wetter conditions surprisingly. And sure enough, back there in September and the mossies were just about — actually I was there a little later as well in November. In November the mossies were just about carrying me away because of the great conditions. They had warm and wet and it was growing like no tomorrow. But certainly yes, the continuation, the forecast for the winter was for wet. And even though in Birchip itself they didn’t get a huge amount of rainfall, it certainly was a wet pattern overall.

So like I’ve said, we’re still looking at the odds but we actually had reasonable confidence. So probably the skill map was at the low end I would say of what we expected to happen in terms of our confidence level, because those climate drivers were looking pretty good.

So moving onto this year, to give you a quick round up and quick outlook as well for this year, look so far we’ve actually had a very —well we had a very warm winter. Overall for Australia believe it or not it was our warmest winter on record. And not just by a little bit but quite a long way actually, we beat the old record by 0.3. And the old record was 1.0 plus 1.6, we got plus 1.9. To a climatologist that’s an absolute shellacking.

Bit cooler in Victoria, particularly in the nights and so on when we had quite cool conditions but clear skies. Now why was that? Well in part, maybe it was a little bit to do with it bumping up a little bit towards El Nino. It was also a bit to do with warm conditions we had in the Indian Ocean as well.

But looking forward now, so what we’re seeing at the moment is again we’re looking at this box. This is a box we use as a proxy for El Nino and La Nina. Looking at the box out there it’s starting to actually cool off a little bit. It’s starting to cool off, head away from what would normally drive us towards dry conditions. And, most of the models are pushing a little bit on that cooler side, so again heading slightly towards that wetter side of the chart as we go into the months ahead.

And in the Indian Ocean, the Indian Ocean Dipole that we heard before, it’s really just the difference between the east and the west of the Indian Ocean. And in some ways this is quite a critical pattern for us. It’s also a very critical pattern for Horn of Africa and indeed some of the terrible droughts we’ve had there it’s been when they’ve had cool water off the Horn of Africa.

So at the moment it’s quite the opposite though. This Indian Ocean pattern is pushing us slightly towards the drier side over the next few months. So it’s pushing us towards something that may give us drier conditions. So you’ve got a wetter push, slightly wetter push from the Pacific, a slightly drier push from the Indian Ocean.

So our outlooks, well first of all for temperatures though, we’ve got lots of warm water around Australia which is certainly bumping things up in the north, got some good skill at this time of year. We’ve also got some very warm water down here in the Tasman, and it’s keeping things warm for Victoria and Tassie as well.

We’re slightly worried about heatwaves as we get into this spring and into summer. There’s certainly high odds of having excessive heat, or a number of days in the top deciles or the top 10% of recorded temperatures, and likewise down here in Victoria and Tassie as well. So we’re a bit worried about the heat and that’s why you might have heard a bit about the — slightly on the dire side bushfire outlooks at the moment. We’ve certainly seen some heat already.

Some 12th of September, what is it 14th today, it’s two days ago Mildura had Victoria’s first 30 of the year and actually slightly earlier than what we’d typically see. A couple of weeks earlier than what we’d typically see. Sydney had its near 34 yesterday, and actually yesterday we also saw Australia’s first 40 of the season up in Wyndham, so all happening at least slightly earlier than normal.

So part of the reason we’re a little — another reason why we’re a little worried about higher temperatures this summer. And you’ve been hearing about the energy issues, of course the energy people are quite worried about high temperatures too.

So this was the spring outlook. Remember we said something was pushing us slightly — slightly wetter; some was pushing us slightly drier. And that’s certainly what we see with the outlook, looking a little bit drier out here in WA for spring, possibly a little wetter in southeast Queensland, northeast NSW. But, I’ll check my watch.

Okay I can show you the next slide. So this was issued late this morning.

So now we’re doing two outlooks a month. So I’ll just mention at the final slide. But just this morning we issued our first October to December outlook.

And those two competing influences, the slightly wetter pattern from the Pacific, the slightly drier pattern from the Indian Ocean are certainly having a battle royale and keeping things very benign in terms of probabilities for rainfall over Australia. So in other words, if no strong push towards exceptionally wet or exceptionally dry, it doesn’t mean you automatically bet on average. You are probably better to go and look at what sort of your typical range of rainfall at this time of year.

Then look at things like soil moisture and the other factors, the temperature outlook and so on to make a decision about where to go. You can see with temperatures still looking warm particularly in the southeast and across the northwest half of Australia.

As I just said we’re now issuing our outlooks twice a month. Eventually we’ll be going to more regular outlooks again. But this is only the second time the Bureau started — since it started doing the seasonal forecast back in 1989, so it’s only the second time we’ve done two outlooks in a month. And this one will be updated, I think on September the — September 28th. So keep an eye out for that.

But the ones at the end of the month are generally going to be the ones I’d really trust. It’s a bit like a seven day forecast. At day five, if you’re wanting the forecast for the footy on Saturday, you’ll have a look at it. Oh, might be wet, but you really have a close look on the Friday and take that as the main forecast, a bit like saying to you at the end of the month it will be the best.

So just in summary there, a warm and dry winter and an early start to the bushfire season. As many of you have seen now, that was generally the forecast for large parts of Australia too.

The ENSO and IOD are neutral and we had a warm spring is favoured in the north and southeast as well. Looking a little bit drier in the southwest but that outlook that was issued an hour or two ago is suggesting that those odds are backing off.

If you what to have a bit more, and I believe the slides will be circulated later, if you want to have a bit more of an education, or a bit more learning around how to use the outlooks, we’ve put together a training module. It’s free to do.

We did this with University Centre of Atmospheric Research in Boulder, Colorado.

And, there’s some good tools there to help you learn a little more about seasonal outlooks.

And that is it from me.

Page last updated: 05 Jul 2021