Rural Financial Counselling Service (RFCS)
Weather and moisture tools
The very fast breakMonthly three minute clips of the seasonal climate outlook for Victoria or South Australia
Seasonal forecast tools
Forewarned is Forearmed - Climate outlooks forecasting extremes was presented by Dale Grey, Seasonal Risk Agronomist from Agriculture Victoria presented to the Horticulture Industry Network, December 2022.
- Climate models
- Climate outlook
Transcript: Forearmed is Forewarned - Climate outlooks forecasting extremes
So the Forewarn Forearmed project, was a really large project that involved a cast of thousands, I think in horticulture involved viticulture, Wine Australia was part of the funding for that, but it was GRDC and Cotton, sugar, and meat and wool, and grains, and it's like a myriad of people put funding into it. Myself and Graham on the line, were involved with Ag Vic as was people from other department of Ag's around this country as well, working with the Bureau of Meteorology to come up with a whole heap of new products that I think now move Australia from being perhaps the most, I wouldn't say the most poorly serviced by their climate forecast, but certainly the most simple in terms of our above and below median forecast that we have, to now have more data from the access climate model than any other country provides. This model's really been tortured now and providing quite unbelievably rich information from that. So what I'm going to do today is actually go through a bit of theory because we all like a bit of theory. Righto. So this is a mystery location in Australia.
We're looking at some August to October rainfall here. Historically, over 124 years, the average is 125 mil. The medians 123, the lowest in that data set is 46 millimetres, and the highest is 278. And if we rank that data from lowest to highest, it looks like that, which is that same data, but sorted. And if we divide that into 10 separate sections, that's called a decile. Now in the grains industry, deciles are by no means new information. We've been talking about them for as long as I've been in the department, I suppose 27 years maybe. In other industries, they're perhaps less used, less known but we reckon they're a really critical way of discussing climate and rainfall and temperatures because it allows you to compare apples to apples. Literally, there's a nice horticultural reference. Doesn't matter where you are, if you're in a low rainfall zone or a high rainfall zone, if you say, I've had decile 10 rainfall, you've had the rainfall in the highest 10% of your records. So it's quite unusual. If you say, I've had decile one rainfall, it's the rainfall in the lowest 10% of the record. It's pretty dry. And so this is the definition of deciles. If you've ever had to create them yourself in Excel, you've got to rank everything and divide 'em into 10.
So the interesting thing about deciles is a decile is not a number. A decile is a range. So you often, if say for instance, we had rainfall that was 90 millimetres at this location, you'd have a look there. It's between 88 and 104 mls and you go that's decile three rainfall, but there's no numerical number for decile. Some people sometimes take a median point between these two values, but that's erroneous. A decile is a range. It has a lower boundary and a higher boundary. And when you're talking decile one, the lowest boundary is lowest on record. And when you're talking decile 10, the highest boundary is highest on record. So that's how you define deciles. And, you pick a number in the record or the number you've most recently had for November, and you look at where that fits. And then you can choose the decile ranking and describe your data, just not in terms of physical millimetres or degrees, but in terms of temperatures. And you can do this with, median house prices or interest rates over time. Size of fruit over many years of harvest. Whatever you want you, any data set can be ranked from lowest to highest and described in terms of deciles.
In terms of the Forewarned is Forearmed project, it was decided to do things a little bit differently. But you may have seen these terms here in the Bureau's nomenclature. Things like very much below average, above average. This is not just random word generation. These words in bureau nomenclature mean something and so very much below average is the decile one rainfall, and very much above average is decile 10. Below average is decile two and three, and decile eight and nine is the above average and deciles 4, 5, 6, and seven are termed close to the average. And when you rank the data, you can see how that can come about, such that, the decile one is clearly the very low, the decile 10 is very high, two and three. There's not great differences between that. There's not much difference between eight and nine and four, five and six. Seven has a spread, but it's kind of, it's almost the middle third of the data, but it's not perfectly a third. It's a bit more than that. But when you get average rainfall decile four, five and six, seven is not distinctly far apart from itself. So these words are used by the bureau to dis to talk about these decile rankings.
In the forewarns forearm project. What has been done is we've done a bit of a sneaky and we've looked at Quintiles, which is dividing the data into fifths. Terciles are divided into three, quartiles divided into four, quintiles divided into five, deciles into 10 equal portions. But by definition here with the, with this quintile sort of thing, is that quintile one is decile one and two combined and decile nine and 10 is the quintile five. And the important thing with this quintile, is that there's a 20% chance, historically, if you have a number and you're looking for where it fits in the historic record, there's a 20% chance that it's falling somewhere in here.
I should have said before that when you have deciles, there's a 10% chance that your bit of data falls in one of those deciles. Just by definition there's a 10% chance it fits in and with Quintiles, there's a 20% chance that when something happens, it can fall somewhere in amongst here. And so what we are looking in this project is we're looking at whether those odds increase above 20%. So always keep that in mind when we are talking about this stuff, that the 20% is just the boring chance of whatever happening normally. And if you're lower than 20 or higher than 20, then it's a bit more abnormal. Deciles, the quintiles have an upper and lower boundary to their range. And in terms of the nomenclature, the Forewarned Forearmed project is called unusually dry decile one and two, unusually wet nine and 10, and the average is decile five and six. The dryer is decile three and four, and the wetter is decile seven and eight. So from now on, we'll see these sort of words used, and once again, they're not random word generation. They're actually, they have a definition and a meaning as to why we are using them.
So let's go now and look, we won't be looking at a live capture of, we can go perhaps towards the end and show the website live, but to find the products, a lot of people don't, have never gone to the bureau's actual climate page. There's two places you can find it. This is their home screen here. You can go to climate and past weather or down the bottom here, you've got water, climate, environment. And if you hit climate, both of those two sections will take you to the same page. Which is this one! This is the website if you wanted to take it down. You want to write it down if you're looking at the recording. But this is the climate page here. There's a power of information that I and Graeme go to here on this site, but specifically the forecasts are up here in the rainfall outlook. It's the beautiful map there of our above and below median forecast.
If for interest, you're in looking for the rainfall maps for the last week, the last fortnight, the last month, the last year, that's this box just here, which is the recent rainfall one, and you can get the decile maps for rainfall there as well. But we digress.
That brings you to this section here, the chance of above median forecast. And there's a number of tabs down the side here, which are new as opposed to some which are older. So you click on that one. And it's important to remember that for a good year and a half now, I suppose we've been able to get more forecasts than just the forecast for the next three months, which is this sort of one to three month forecast.
So historically, that's all we've had. About a year and a bit ago now, the access model was expanded such that you could now get forecasts out for one week, the next week, the first and second week combined, the second and third week combined, months one and months two and four. So it's not just the three next three months, which is what we see classically on the forecast in the media. There's all this other shorter term information, which is not a weather forecast, and that's the whole thing about Forewarned Forearmed, is it was discovered that the access model and climate models, per se, around the world which initialized with the weather forecast as they see, it on the day they're run, the weather forecast is better going out past seven days. Most weather forecasts are useless after day eight and no better than flipping coins. But the climate forecasts have got some of that embedded weather knowledge, but the overarching climate signal, and they can start having skill into weeks two and three that the weather model is useless at, because of just random chaos in the weather.
And so that's where this project has really shone I think, by teasing out weeks two and three, and particularly looking at the spikes in rainfall and temperature, in dryness, in coldness, that might be useful in managing horticultural business and agricultural businesses in general.
So when you click on the good old, above and below median forecast. This is one from a couple of days ago. We've just chosen the month forecast here of December, but you can choose the weeks here by choosing different tabs or the good old three month forecast. This is the December forecast. This is just the chance of above median one.
Now hopefully not telling people to suck eggs here, but we've got a lot of white on this map at the moment for Victoria, which is meaning that the month of December is, was predicted at that time to be 50% chance of median and people would often determine that meant average rainfall. And it's important and of pains to explain that is not what this forecast means. It means there's a 50% chance of getting greater than average rainfall. But by definition there's a 50% chance of getting less than average rainfall as well. And so this white section here is really prepare for anything, anything could be happening.
The browner forecast that we see up here in the Pilbara and the Kimberley are a dry earth forecast. There's 30 to 35% chance here of not exceeding the median. Sorry. Yes. No, of exceeding the median, so a low chance of getting high rainfall, but a 30% chance of exceeding the median is a 70% chance of not exceeding the median of being drier. So could be wet. There's just greater chance of it being dry. And we've still got a little bit of green, mainly at the top end there. An 80%. Of exceeding the median, but by definition a 20% chance of it being drier than the median as well. So always to remember, we're talking about probabilistic forecasts here, and there's always a flip side to whatever the prediction is. It's not an absolute forecast. It's not what's called a deterministic forecast like we see on weather models, which the weather model is one run once, and the answer is x. That's what it is. Now as we go in future, weather models are going to be spun up many times, and we are going to start seeing probabilistic weather forecasts. We run the model a hundred times, and it's predicting that the rainfall is 80% chance of it being above 20 mil, but there's a 20% chance of it being less than that. At the moment, that's kind of the cutting edge of weather forecasts, of doing lots of it, but all the world's climate models have been run like that for many years now. In terms of running the model a lot of times, getting a lot of possible outcomes and looking at the averages of those.
So the first new product that we had, which is not overly new, that's, it's been out for probably almost three quarters of a year to a year now, was the first of our extreme forecast, which is just the extremes map. So we are looking at the chances of unusually dry rainfall here for the month of December across Australia in a map format and remembering that unusually dry is the decile one and two rainfall or quintile one. And you can see on this little legend over here that the average chance of that, of getting that is 20%, remember. And so we are seeing some areas of Australia here for where there's a 30 to 40 or 40 to 50% chance. Of being in the driest fifth of records. And so the model is clearly, almost doubling the odds there from 20% up to 40%. And that's what's on the side here. You got to double the chance here, if you move from 20 to 40, you're doubling the chances. There's double the model runs that have come out in the driest thing. Now, I should explain that the bureau, when they're doing their forecast, they do this every day and they use the last three days of forecast, their big super computer can run the model 33 times in a day, and every forecast is made up of 99 model runs of the previous three days. We can round that up to a hundred for simplicity. So this forecast here is made up of a hundred separate model runs. You'll see the unusually wet is beside that, and you can see that the prediction for December has, to be honest, this had changed. Now, I don't know what it's done this morning. We might go and have a look, but as of last, I think it would've been last Friday this access ,model had gone from much wetter signals from December to suddenly being sitting on the completely on the fence and going, anything could happen. And that was a major change in the forecast out of this model, which for the better, I would've thought in horticultural land, and in my industry of the grains industry as well, desperately looking for some drier weather, indeed. And it looks like, at least from this model, was seeing the tea leaves differently to the previous week. And that the, the signals had kind of gone off for the wetter December, which was which was great to see.
The other thing you can go to is temperatures. And so obviously for temperature, you've got maximums and minimums. And when you do temperature, you've got, these are the, you've got a toggle between maximum and minimum, and you've got unusually cold and unusually warm maximums. So that's warmer than normal daytime temperatures normally. And then you've got the warmer or cooler than normal minimum temperatures, the night-time temperatures. So this is this looking at a, here, an example here of a two week forecast from the 6th of December to the 19th of December. Looking at the chances of unusually cool. Remembering once again, 20% is considered normal chance, and we've got some increased chances around Esperance of it being much cooler than normal for maximum temperature and along the south, south of the divide and east of the divide in New South Wales. And if we are looking for warmer temperatures, unusually warm ones, same period for the two week period, we can see the top end really screaming up there to 80% chances of being in the warmest fifth of records being decile nine temperatures up there. I presume they're still in the middle of Mango Harvest up there. Maybe coming to the end, but that's probably going to hasten that on pretty quickly too. And I think they've been having some very warm temperatures up there anyway for the last many number of months. And then just sweeping, flipping over, this is now the minimum temperatures should have been. What have we got there and usually cool. Do. Should be the minimum temperatures. I'm hoping it is. Chance of Unusual night, unusual nights 13th. This is moving to the 13th, to the 19th period here now. And we're just seeing same data really. I'm not convinced that is actually, I reckon that's maximum temperature. Actually. I think it was supposed to be minimums. If you hit minimums though, you get the same sort of thing going on there. But this was for the second week out in the forecast from the 13th to the nineteenth. I might have made a booboo there in my rush to get things done because this, that should be highlighted down here, minimum temperature and we get the same choosing things there as well. The other things you've got here as well is you can just see what the average temperature is for that fortnight. You can do that for rainfall. You can see what the anomalies' predicted to be. And you can look at the accuracy as well. You can look at the skill for predicting for those two weeks and one month periods for any, anything you're looking at here, you can look at the skill of the model historically there as well.
Alright, which brings us to a new product or another newish product here. This has been around for a little while as well. This is the one that Graeme and I have loved the most because this is a point and shoot thing here. You can, when you get to that chance of median map, you may or may not know that if you've accidentally just clicked on it, you'll get a drop pin appearing and you'll get this little bar chart coming up, which is the decile bar or the Quintile bar chart. And this is the one for 13th to 19th of December, and we're looking at the chances of unusually, not necessarily unusually wet. What we're seeing here is the spread of the most recent 99 model runs, and we can see that there's a close to 18% chance of being in this decile one to two. There's a, that's here ,18% chance there's a 28% chance of being in the unusually wet. So some increased chances there of being wetter, but not strongly skewed, to be honest. Many of the forecasts we've seen in the previous three months, have had 3% chances of being decile one and two and 10% chances of decile three, four, and 40% chances of being decile nine 10. They've been really strongly skewed towards the wetter end, but this one's a bit more blander and a bit more boring at the moment, which is a good thing to see. There's increased, slightly increased chances of wetter but, you know, not distinctly different across the rest of the decile ranges. Why this is so cool is that this is now brings us out of the archives of being plus or minus the median to now seeing what's the chance of it being drier? What's the chance of it being wetter, and for my money, most excitingly, what's the chance of it being close to normal? Which we've just never had before. So now you can see there, slightly decreased odds of it being more around the normal there, but a 59% chance of being above the median. But this is really all the model runs of those 99 model runs that make up that plus or minus median forecast of 59%, giving you much more information there and you can click on that. This is just the week one, but you can look at months or you can do this for the three month outlook as well. And you can, I think we might have a crack at, we can do this for, I think we've just brought that in a bit more. Just brought that in and zoomed that in, I suppose, looking at that the historical median. For, this is for Windermere too, so near Ballarat, this forecast, so these forecasts are location specific and, and you either, yeah, you either click on the map or I think, didn't show before, but what you can do is you can simply type it in. So the little question, a little magnifying glass here, you click on that, you can type in whatever location you want and it pops up. And that'll bring up the grided forecast from the access model for your actual location. And bring up the what, the decile bar chart for there. I've explained what these bits here are. You can see at the top of this forecast there are three little icons. They're important because this is the decile bar chart icon. Icon. This one is the climogram, and this is the probability of exceedance graph, which we're going to see soon. But these icons up the top here allow you to flip between three new products in terms of how to explain this or present this data.
The stars on the side represent the skill. So if you've got one star, there's not much skill at all. In fact, flipping coins is just as good or better. If you've got two stars, you have some skill, certainly better than flipping coins. If you've got three stars that's pretty high skill. So that's really just the simple way of looking at that. If you want to go into the statistics of how that's worked out you can, the little information button here that you see around provides extra details. You do, you can, all these things are hoverable as well. You can hover over these images and things will be popping up that can take you off to information sources. Just to show there that the, when we are comparing the skill and the averages of the data, it's from the period 1981 to 2018. It's a 30 year period of history because, and the reason for that is when you run the model back in time for every week back to 1981 and you've set it up with the data that existed at that time, that is very computationally expensive. It takes a lot of computer grunt and time to run the computer models and to derive that data. It'd be great to go back to the fifties, but you simply don't have the data back then, the satellite data to initialize the computer to do it. So that's as good as we've got in terms of the averages and the skill of the model. But that's, you know, compared to other models in the world, that's about as good as it gets as well.
Here's just an example of temperature. We're going to Gippsland to Lindenow to the veggies there, Danielle and the other three AUSVEG people too. This is just looking at minimum temperature forecasts. A quintile bar chart there showing that there's a decrease, half the chance of it being, 12% in fact of being unusually warm. It tells you what unusually warm would actually be. So greater than 13 half degrees for minimums would be unusually warm. And the chances of decile 1, 2, 3, 4, 5, 6 are pretty much sitting on the fence and seven, eight for that matter too, so slightly decreased chances of really warm, but pretty much equal chances of everything else.
Rightio, which brings us now to a, just a little bit of theory around a different product, which is the climogram and the theory behind a box and whisker plot, which is interesting. I'm a bit old. I was in, did high school in the eighties and box and whisker plots did not exist in my mathematical knowledge. It didn't exist. But they simply exist of a box on a graph and lines above it which represent the whiskers. And the box and the whiskers have a definition. I think the box and whisker plots were invented by a statistician around 1974, and they've taken quite a while to come into common parlance, but they are more so now, and people that went to high school through the nineties and the two thousands usually they understand these things. They get taught them. But the definition of box and whisker, is the box is the middle 50% of the data and the median or the middle value is the line across. The top 25% of the data is in the top whisker, and the bottom 25% is the bottom whisker. And the lowest extreme. And the highest extreme are the abound. So what this graph shows is the, is all the data. There's, warts and all that's showing the highest and lowest and the middle chunk and the median. You can get a lot of information out of looking at a box and whisker plot. In terms of our Forewarn Forearmed products, there's just a slight change to that definition, and at the bottom one is just looking at the 10th percentile. So decile one and the top here is looking at decile nine. And the reason we've had to do that is because we're talking about of a model, sometimes the model being a model goes awry and you get some really weirdly dry and weirdly wet answers out of the model, which aren't probably plausible. They're just the model being a bit funny. So that we don't get some really weird spread outside of the definition of an extreme, we've cut down to that 10th and 90th percentile. And so this is the new product you get. And you get that by clicking on this icon of the forecast. And this is really cool because this is either a four week forecast out or a five month forecast. I am drifting towards this being one of my favorite ways of presenting climate data just because it shows either in weeks or months, the four weeks out or the five months, and there's no other product from the bureau that you can see the individual months. And it's, that becomes really interesting because this is the forecast here out for the four weeks from last Friday showing a reasonably tight forecast there for not much rain. And we are looking for here. Once again, this is a location of specific forecast. The forecast out from the sixth to the 13th had arguably quite a widespread coming up in the next week. It'd be interesting to see whether that has changed. Then it was looking out to the next week, which was the middle data here is showing it's probably going to be drier, but there's some really high tails up here suggesting that there are model runs there, sniffing wetter in the wind. And that's really, a real thing that's going on with all the forecasts at the moment, is they're coming back to the normal thing, but those really warm oceans to the north of us suggesting that if you got the right setup, high rainfall is still possible. If we look at the forecast out to five months, we can see this is the, we've got to November, it was decile 10 in a lot of locations in the top bar quintile, decile nine 10, and the colour schemes here, this is the history here of 9, 10, 7, 8 four five and six for the average. And three, four in the bone and one, two in the brown. The colour schemes across products are all interrelational. They're all being selected for a purpose and they all kind of mean the same thing. But we can see that the December forecast there sitting on average. The January, February, and March forecast all sitting around average as well, which up at Mildura is not very wet. That's, 10 or 12 ml per month. Once you get to summer, it's not very exciting rainfall and people will say, bring it on. We can, that's not going to lead to a powdery blowout at that kind of rainfall. But if we were getting sixties and seventies, you can see that there are model runs that are up there quite high. So it's just not saying, 25% chances up here in this whisker of being much wetter, but the medians are quite low. But it's kind of a low bargy, it's trending wetter more than anything occasionally. So I'd be thinking wettings are, wet is a possibility, but I'm expecting it not to be that wet. But this is what's great about this spread is it's allowing you for your business in industry to look at that and make your own interpretation about what that means for your management.
This is the temperature ones. You have the same thing. This is the light blue, which is the decile one, two bottom quintile temperatures. This is nine 10. Earlier on in the month, we were looking at the chances of it not being, you know, there were cool temperatures, particularly drifting through it into December, but that looks like that's now kind of sitting on the fence more, and this is for maximums and they're kind of sitting on average for the weeks ahead. If we look at the month forecasts, they, to all intents and purposes in Victoria, look cooler to sitting on the fence for December and then drifting warmer into January, February, March. The excess model does have things kind of going warmer once we get into the new year. This, I think I could have had a minimum temperature forecast there as well, but I've chosen for brevity not to do that.
Which brings us to the final product, which is the second final product, the probability of exceedance graph. Now this is really for the nerds amongst us. I'm a bit of a nerd, but I probably don't have a lot of need for a probability of exceedance graph, but hydrologists amongst us, this is what they use all the time when they're designing dams or they're looking at river flow. They're talking about one in a hundred years or one in 500, and you go what a crock, we don't have 500 years of data. But there, that sort of stuff is constructed from these probability of exceedance graphs where you've got the actual precipitation or temperature or the value you're looking at, costing millions of dollars here. The insurance industry uses these things all the time when they're working at your premiums. You've got an event occurring and the chance of that occurring and a graph constructed. So if you click on this icon, on the little graphs, you get the probability of exceedance graph. This is for the grape growers at Lilydale for the week of six to or fortnight from six to 19. And you get a red graph, which is the climatological forecast. It's historically always happened, and the blue graph is the actual forecast. Now for this forecast, we can see there is stuff all difference between the two, which is suggesting very much a sitting on the fence forecast for that period of time at Lilydale. But I can tell you that the forecast from this model for probability of exceedance graphs for the previous three months, that blue graph has always been sitting outside and above the historical forecast, meaning that there were much greater chances of getting higher rainfall from the forecast. Depending on I, I think this will be interesting what this product gets used for, and the more people that look at it who are into data, people looking at sort of matrices or risk and reward of the probabilities of certain events happening, can build their own kind of data in Excel I'd imagine, by looking at this model forecast and punching in the various numbers for the different chances of things happening, and doing that. So as is your wants, if you're right, that sort of stuff you can go your hardest there. And this is showing all the 99 model runs that have come out, warts and all and, doing what you want with that. I think I've, you know how that's probably telling you to suck eggs, how you read these graphs. You can read them from this way, but you can also read them from that way. So if you go this way here, what's the chance of me receiving around 15 mil for this period? It's about an 80% chance supposedly there of exceeding that, given where in Lilydale, that's a relatively high rainfall location. What's the median chance of rainfall? It's a 50% chance of getting around, whatever's written up here, 28 mil. What's the chance of me getting 45 mil? It's a, it's about a 30% chance. So you can read every sort of combination of this.
The last product we've got is back in those original maps. It's a map forecast, it's down here. This is where we were getting those extremes maps that you, you don't have to, but you can click anywhere on them to get those quintile bars and the probability of exceedance and the climograms up. But this is just a map showing what's the chance of getting a user defined amount of rainfall of 15, 25, 50 or 75 mls in the next three days, that's all we are looking at. And you can look at a three day total over weeks. This doesn't go for months. Now, this product was asked for by the Northern Zone people. I didn't explain that we had a lot of user groups, some of you people might have been involved in that user group stuff for viticulture. But we had user groups for beef and sugar and, groups of industry people, farmers and other industry agronomists and consultants who were providing information into this project to frame the way these things looked and what the, what we were asking of the access model to, initially. And this is one that the Northern people asked for. I'm not quite sure what purpose it's going to have in the south. I think we are going to find out. But it's classically used for the northern wet season bursts of the monsoon. But down south it might be useful for picking and crop protection where you're looking at certain amounts of rainfall potentially bringing you unstuck over a certain period of time. In, in grains for interest, we might be interested in this for, the timing of the break. What's going to happen there when we are getting the start of the season in, March, April, May, heaven forbid, June or July. But we will see, but think time will tell as to what this project, what that actual graph is useful for. But you might be sitting there yourself and thinking initially right now what you might be able to use that for.
We'll take you through just a live example. See if, we'll see if that forecast has changed for December for interest's sake. Righto. So this is the Bureau's homepage. As I said, you scroll down, here's the climate and past weather. Here's the climate, one of the, either of them will take you there. There's the forecast, the long range forecast. We'll click on that. Up it comes -chance of above median. There it is. Here's the choice of weeks. We can go week, we can go the next week looking like it's back to normal for there. We can go to month. So here's our forecast for December. It doesn't look like it's changed dramatically by the looks of that. But give me a location, Mark in Victoria.
Oh, I'm going to be biased and go to Goulburn. I'm down the southern end, so Buxton please.
I know what's going to happen here. . My, alright, Buxton. There we go. Buxton Victoria. Please wait. Righto. So there's our decile chart there for Buxton. Not showing much of a difference there for December, pretty much the odds of anything happening are all over the shop. So I think that's what's interesting is this tells you that there's no overarching climate driver at the moment to necessarily be spinning it up wetter or dryer. And that's important information to know. Like it's not, the chance of average is normal too, but the chance of really dry and the chance of really wet is the same as well. and that really is a plan for anything forecast which is better than planning for average and being blindsided by something really dry or really wet. And people would go, oh, the Bureau didn't forecast that. In actual fact, the odds were the same. There would've been there's 20 model runs that are coming out really wet out of that model at the moment, but there's 20 coming out really dry as well. If we go to the, this is the little button up here. We'll go to the, somehow gone to Taggerty now. But anyway, that's must be the closest you get to Buxton. Yes. This is that four week forecast showing the spread and that's showing, these are really wide, box and whiskers. There's nothing tight about that prediction there, or the ones afterwards showing that for the weeks ahead, the model is all over the shop. Which is, I think indicative of this breakdown of this La Nina period. It's, things are, oceans really warm to the north of us, and anything is going to be possible. If we still get a connection up to the tropics, the game is on, as indicated by some of these really wet things here. But the actual medians of these things are showing there's a chance of a wetter week from the 10th onwards and then coming back to normal odds of rainfall in the weeks leading up to Christmas. And then you've got the five month forecast here. it comes back to normality. I'm hovering over here and things are popping up in terms of the numbers that define the different parts of that box and whisker. And then if we get so excited, we can look at our, let's go for the three month one, for instance. Let's see if there's a difference in that. Come on down here. Got to click back on the little button there. Is that happening? Here it goes. Here's our probability of exceedance forecast at Taggerty. And you can see there is a little discrepancy there between history and the forecast being, it's sniffing a little bit wetter in the wind there for the next three month forecast. I think that's because I think when we go to the one month one, I think January sniffs out, yeah, so January is predicted to be a little bit wetter just from the median forecast. So December's coming back to bit to normality, but then January's sniffing a little bit more wetter in the wind. There's the forecast. Okay? Yep. There's the forecast for January. You can see that, the chance of decile one, two is down, but nine and ten's not blowing out. It's kind of being pushed up by decile seven and. There more so than other, it's almost a bell shaped curve there towards median to wetter. But I'd be at pains to just explain that, that these graphs are just a world apart from what we saw during those wet times where this model was predicting things pretty well as looking extreme, and in fact they turned out to be.