 So we're all on the same page and thanks. Thanks really for your interest in this school So, you know, we have had a couple of trainings Already we had one at the apex climate center. I think the same was there about about a year ago and we had it we had a little mini training a couple of weeks back at the the ASEAN climate outlook forum precoff that the Ryzan Organized but this is the first time that There's a training course on actually how to use the database So this is this is really really a first and we really appreciate you being the guinea pigs and and and you know being being engaged and enthusiastic to to get get your hands on on the on the data and See what we can do with it. So This is week two and I'm I'm really looking forward to hearing about what happened in week one and and what every what everyone's doing How the practical exercises how the practical two exercises have gone So in the week two as Adrian said there's quite a bit more focus on more the application side But also we have we have some dynamical modeling we have Steve here We didn't want to make it all just you know Dynamics in the first week and an applications in the second we wanted to we wanted to mix it up So we've got Steve will know here who's going to be talking about modeling the MJO and also the the stratosphere as being another of the sources of predictability on the S2S scale But today we'll have it'll be mostly sort of applications oriented and I'll I'll give the introductory talk on some On applications and then we'll have Frederick the other Frederick from ECMWF here Who's gonna who's gonna talk about? applications in more detail in drought forecasting today and Then tomorrow will be on on the other on the flood on flood forecasting and actually have a there's a there's a flood forecasting game They're all looking forward to on on Tuesday afternoon And then we we have we have Vincent Moran This morning and tomorrow who's going to be talking more about the downscaling and weather climate relationships especially weather types today and then Towards thinking about a prediction of local scale rainfall tomorrow and then through through the week On on when so we'll be having the the practical exercises in the afternoons And so I think the thinking is that this week There they'll they'll be more on on your case studies and and we can discuss what more that that much more and as Parola said that we may have a also this afternoon a Demo on how to install all the software on on your own laptops And I thought well, maybe this afternoon might be a good time to have that after the the verification There's a talk. There's there'll be a talk on verification. It won't be 90 minutes We're probably more like 45 and that might be a good time Then Ryzan on Wednesday morning Ryzan is going to describe this this initiative with with asian cough I think it's it's a nice. It's a really nice example. So I'm pleased that Ryzan volunteered to give give a give a talk on that on Wednesday. And then on Thursday we have some room for Participants presentations, and I don't know if if you've talked about this at all last week But we'll play that by ear. I know there's a couple of people who'd like to present something But there's there's room for there's there's plenty of room in the schedule as we go later into the week so That's sort of what the schedule looks like any any questions about any questions about the program for this week So if not, we can we can get right into it So I called this I I revise the title of this to be applications consideration So this is a huge topic and what I wanted to basically do here is just to introduce some of the some of the considerations on the application side The types of Forecast user and application and then what makes forecast information valuable to a user What do we need to be thinking about things like the salience of information? It's timeliness How credible the information is Is it understandable and is it does it come from a legitimate source? And the basic question is here is can we translate scientific information into into useful knowledge? It might seem obvious, but I was just I was just sort of Skimming through this this book. I don't know if any of you have seen this. It's a book by a guy called Nate Silver And it has a it has a great title The signal and the noise why so many predictions fail, but some don't One of the things he's saying there is things like trying to predict the stock market has been a miserable failure or trying to predict Recessions and things like that But some things are actually more successful and one of the one of the things that he says doesn't fail is that his example is It's weather forecasting so We will be seeing Thinking about that you can think about that in the in the case of I hope that s2s is also going to fall into this This category of ones that don't Adrian Yes, right So in the context of you know why why this is challenging to you know Produce information and have a use this is a One illustration that he gives and what's shown here is the the global GDP per capita in the world and showing the evolution of this and he pointed out the the invention of the printing press in 1440 as being a seminal event where Suddenly, you know information could be shared because before that, you know everything had to be copied by hand but that The GDP stayed more or less flat for another two or three hundred years and the steam engine was invented in 1775 and then it took off so there's this huge lag between you know Information being starting to be shared and produced and and it being used in the history of history of GDP So I just put it up for fun and saying well, I hope it doesn't take that long, you know for s2s forecasting capabilities to To to have to have some real value But just to underline that it's not it's not a trivial thing to if you have information That it really be translated into something of value and this is something we've been struggling with at Sorry at at the IRI for almost 20 years now and So often you'll see well applications That's just you know a little box at the end. Okay, we'll we'll we'll make the forecast and and everything We'll do all the science and then at the end. They'll be the applications and The s2s project is is maybe a little bit like that in that all the all the people in most of the people involved in In that are on the modeling And the science side, but there actually isn't that much On on the application side yet. So that's what we're hoping that you guys can help I don't actually have that schematic, but maybe it was shown last week the the s2s graphic I'm sure that Frederick showed that where you have all the sub projects and then you have those pillars and then you have the database underneath and Then two of them two main two pillars with all the all we have all the the science issues About predictability and then and then all the modeling modeling issues things like initialization ensembles, etc And then the last one says needs and applications and then basically it says at the bottom Well, we will liaise with the the world weather research programs Science societal economic research Sarah and applications project and we'll somehow get this done. So So far there's not been much on the application side. It's something that we really want to we want to show progress on and the WMO Parallel Rootie actually the WMO the the the world weather research program chief will be here on on Thursday and Friday and he's been Hammering on us that we need to we we need to create some what he calls exemplars which which highlight real Application potential application value of s2s so that we we can present this at the next executive council of the WMO and Really get all the members and and members of the WMO Met service is really interested in this and we know from from you guys and and from Talking with people in in in Met services around the world that there's a there's a huge amount of Interest in in s2s and so This is just the beginning that the databases is now online Fresh and so this is we we just have to now Show what we can do with it Okay, so with that I'll I'll get on with the rest of the talk So what about the types of for class user and application that that we could think about for In the s2s context So I just basically Delineated two major types of User and user and application that I could see for s2s and the first would be I think really the obvious one That this is hazard early warning and enhancing preparedness to high-impact weather events And that was what the s2s was is meant to be all about In the in the mission statement of s2s that Frederick showed last week Then it says you know with special emphasis on high-impact weather events So the thinking is that hazard early warning that's something We have hazard early warning from weather forecasts That's that's something we can push out into the sub seasonal range another one that's perhaps not quite so obvious is Things like management decisions in weather sensitive operations. So even if it's not You know a flood or a drought there may there may be management decisions that make can make use and make make use of sub seasonal forecast and to optimize optimize their their their operations so I mean there's a huge range of range of users from from very sophisticated and And I'm sure a Frederick or maybe Frederick will say something about some of these specific users in flood and drought In in the in the talks coming up But you're really ranging from things things like reservoir management sophisticated operations for hydropower say in Sweden down to Developing country use users and I want to sort of emphasize that these ones here in the talk And this was mentioned in the program that I would say something something about GFCS But this is where the WMO hope WMO is Hoping that S2S can can really make a contribution to S2 to W to to GFCS and The most obvious is the way I'm sure that most of you already seen this schematic of GFCS and the way that is S2S is is is housed is mostly down here in research modeling and prediction Pillars of this As I've sort of been mentioning going along But there is also this this aspect of the user interface and climate information system So we want to be able to really contribute to this so that we can inform Users government private sector research agriculture water health et cetera et cetera and in fact these priority areas of GFCS agriculture food security Disaster risk reduction health and water I heard some of those being mentioned by people when they were introducing their interests here and These are where we're hoping also that in your in your projects through the week We will start to sort of explore the potential value of S2S Forecast information and linking in your regions with considerations of agriculture food security Disasters floods things like that or your health considerations or Water management so this is a nice little Infographic that I that I like to show that to really starts to get to the question of How how weather and climate information Is being used or has the potential to to be used and what are the types of that information? How is it disseminated to users and and what are the kind of decisions that could be impacted and This is in the case of for how farmers around the world are making decisions based on weather and climate information and This this this schematic work this this view graph with here Infographic was made by by the sea caps, which is the the CGI AR the the consultative group for international agricultural research is climate change agriculture and food security program See cash, so there's a terrible acronym within an acronym. If you know, I mean CGI are Within this acronym and what what it's Pointing out is really I think nicely illustrating that you really need information across time scales when it comes to managing Climate-related with risks, so they're they're emphasizing, you know weather days to weeks Climate variability from months to years and then then climate change as well from from from decades or decades or longer But let's focus here on on the first two which are I think sort of nicely illustrating, you know s to s to s bridging across these these two so Maybe this schematic was made before before s to s so we have we have the weather on the one hand and the the seasonal months to years climate on the other and The s to s is like like you've heard many times Hoping to fill the gap between these two and and link link weather and climate and I think also you'll see through Vincent Moran's talks in the week how how Methods that we can really go about doing that so what what are these types of information and What what kind of decisions? Have they pretend have they the potential to inform so let let's start with the seasonal ones say a Seasonal forecast would be in terms of seasonal rainfall or temperature temperature conditions typically and What could this be used for or it could be used for selecting crops of varieties for for planting livestock stocking rates for example feeding strategies Maybe even the labor market for for thinking about a few if there's a if the the rainy season expected to be Good you may may want more may want to engage more farm workers or intensify crops or diversify them if the forecast is is rather for below average rainfall and And How could these how could this information be? better How do we go about thinking about how to you know tailor the information and seasonal forecast to make it more applicable to these needs Thinking about seasonal climate variables targeted to particular and agricultural risks dry spells or Radies rainy season start date, etc. And this is starting to think about Well, how do we how do we tailor information from our forecast to make it of more valuable more value to to users? Well, what are the methods for going about that? We'll have a lecture on Wednesday on on that specifically on that tailoring Tailoring aspect and thinking about well, how can we if we have sub-seasonal forecasts? Well, how should they be? How should we extract the information that they have? And you can see that if we're getting down to the more daily scare things like dry spells or raising rainy season start date in a Sub-seasonal forecast we could well have more information More specific information on on those then you could get from the seasonal forecast So the other important thing here is for thinking about the applications is that it's the piece in the middle And that's the the vehicles for delivering the the information which is which is absolutely critical and That we've seen that I think everyone who's worked on on this in this area of applying forecast for For applications as has come down to the the criticality of this piece in the middle Here they've emphasized here workshops with with experts as being one possible way Could be training events like this as well or conversation with our agricultural extent Extension agents is working here is also mentioned here And the one thing I like to stress is is the importance of intermediaries in linking with linking these two communities together The user community and the the forecasting community So it may not be the best strategy for these to talk to each other Directly they may not be able to and having the right intermediaries To to as vehicles for Delivering the information is really a critical piece and this really has all to be to be mapped out for S2S So for the weather scale Then we have things like daily forecast up to a week ahead and this could be for things like timing of planting and harvest or timing of a fertilizer application or hazard early warning protecting lives and these are more maybe more you could they're sometimes called tactical decisions and Timing of fertilizer and pesticide application. That is something I think where Subseasonal forecasting also has a has a role to play In in these more tactical decisions So in terms of the what crop to plant you might still use a seasonal forecast But then in figuring out when how you should schedule your irrigation or or fertilizer or pesticide application When that should actually happen then using weather forecast or subseasonal forecast they can come in there and The vehicles for delivering these things are maybe different as well between these two. So SMS messages or radio and television typically for for for weather forecasts. So what should it be? For subseasonal forecast, what is the way that these these need to be disseminated? Okay, so I've sort of introduced in that in that previous slide some of the issues in connecting climate forecast information with users and some of the The ways the ways that people think about doing that connecting now what I want to talk about is some issues of achieving value For applications using using forecast information. So some of the challenges To achieving seasonal forecast value have been identified as salience Need to meet a user's needs Credibility of the forecast understandability of the forecast and Legitimate they need to come from a Trusted source. So I'll just go through those and try to illustrate what these mean and There you'll see there often there's quite a bit of overlap in in In in what these are If it's salient information, then it needs to be understandable and credible for it to be salient for example So some aspects of salience Information should be that it should be specific to a user's needs and and and timely And so one could think that there is is somewhere already where if we may have more we should have more specific information Coming from a sub-seasonal forecast because the lead time shorter than we do for a seasonal forecast So there's there's an opportunity there time timely information According to the user's decisions a sub-seasonal forecast With a shorter lead could be more timely to a specific decision like the applicant the application of fertilizer, for example And then the other one that's that's mentioned here is that the the forecast should where possible address a decision Decision-relevant variables and in seasonal forecasting People have talked about characteristics of local daily rainfall monsoon offset date or could be river flow or drought and this is the issue of tailoring And I thought I just flashed this one up here that some of you may see me present this one before But the relevance of daily rainfall for agriculture I think this is a slide from from Andy Chalanoor and it's just showing for a place in India Gujarat in India Two different years. So these are time series of rainfall through 19th the growing season the monsoon season the summer of 1975 at the top and 1981 at the bottom and actually the total rainfall for those two years was more or less the same So if you had a seasonal forecast and it was perfect Well, then you would have predicted the same same seasonal forecast for both years But if you look at the yields of peanuts for those locations, you can see that they are actually very different between those years in 1975 where we had the more The more uniform rainfall across the season that there was a much larger yield than there was in 1981 Where we had this sporadic rainfall and the rainfall all came at once so obviously this is something well known that dry spells are a problem for for crops and we want to be more specific about About What the information that we can give and within a seasonal forecast? Maybe we can say something about? rainfall frequency the number of rainy days Rather than the seasonal rainfall There's been work on that or on a sub-seasonal timescale. Maybe we could actually say something about these tough the timing of these events They may not be they may not have any predictability. They may be completely random They may they or they may be they may be associated with sub-seasonal modes such as the the man Julian oscillation And here's the other one. I just wanted to flash up and this is monsoon onset date and rice planting area in in Dramayu Which is in Java, Indonesia, and this is a slide from Rizali bore from Bogor Agri-Cultural University and What it's showing here is in the pink bars. These are the Seasonality of rainfall. So this is the monsoon season starts at the end of the end of the calendar year. So we just had the We we just we just had the climate outlook forum in in Singapore For the December January February February season, for example, and then what's what's superimposed on this is for different years 97 98 it's the red 98 99 is the magenta and it's showing the rice planting So they have two rice crops. They have a double planting of rice and then have a fallow in in the dry season And so you can see these two peaks but what's interesting is that The timing of these varies from year to year and particularly this 97 98 event We had where we had the big El Nino. There was a big delay in the planting So you can see that it the planting didn't start on time it got delayed and Then what happened was even the second planting was delayed because of that everything got shunted later And that was particularly a big problem at the end of the rainy season Because they have enough they have enough rain actually in in Java for their for their crop Even if they plant late so they wait for the monsoon to start and then they plant their crop And it's no problem for the first one because this plant is it rains plenty there But then they have problems because everything is late They'll get into the dry season at the end of the second planting. That's a big problem for drought So that if you could somehow accelerate this planting If you if you had a forecast that the that the onset was going to be delayed Then you could take various measures to accelerate accelerate that planting such as making sure the All the seedlings are ready ready for transplanting and things like that so This is somewhere where Seasonal forecasting clearly has a role to play because the El Nino has The the timing of the El Nino it is such that we have forecasts of El Nino In in the boreal summer so they could it they could inform this This this application and the farmers here The in in in Nindramayu province This this is I should say I mean they do have irrigation as well But there are some people with less irrigation and that's where that's where this the climate information could be most most valuable But so it'll be interesting to see how this plays out this year. I Email to Rizaldi to say well what you know, it's it's now been It's it's it's now been almost 20 years since it's 97 98 Event what what has happened in Indonesia? since then We've we've been working with Rizaldi's group for a long time and so what have they done this year that? Well, we'll help them to prevent prevent this problem in In in in 2016 and and he told me that well now that the farmers do through agricultural extension get Forecasts of of of rainy season onset through the through the Met service in Indonesia and he said that that many that there's been much much discussion and He said it's as usual. It's complicated There are many players in this and it will be interesting to see what see what actually happens So the last thing I want to mention here in terms of timeliness Is that is the cropping calendar? This is something that also in Indonesia They've been they the Ministry of Agriculture puts a lot of effort into developing a tool for for helping farmers with their cropping calendar and what Rizaldi bore and the BMKG the the Met service are doing is to to research to what extent can can we help use forecasts to inform that cropping calendar and so here you see this is sort of the uncertainty in in in the yield at the end of the year and You can see that it is decreasing as time goes on But what I want to draw your attention to is that the various stages of the planting that the flowering date and the harvest and where a seasonal forecast could come in at the beginning but then Subseasonal forecast could really be used in this interim for for helping in the timing of these Excuse me tactical decisions and then I This is a little schematic that originally came from Tony Barnston at the IRI that I always like to show in these these kind of events and So I thought I since I wasn't here last week I thought I'd just show it now in terms of the the forecast lead times and where where the S2S is in this This central zone here and the way that Tony had drawn this before was This forecast skill dropping off from the atmospheric initial conditions in the weather forecast and then this ramping up of the seasonal forecast Influence essentially of SST boundary conditions But what he had in the middle here potential sub sub seasonal predictability from MJO land surface, etc. Just something Really really is a hole here Very poor predictability Predictability desert, maybe that's a term we've also heard used and this is this time range where we're hoping to make we're hoping to make progress and As per the previous slide Demonstrate how that could be used in in in user Decisions in different sectors. This is Toward the same same idea here of Multiple multiple times scales and where S2S could sit It's it's something that was developed by the International Federation of the Red Cross Red Cross Red Cross and Red Crescent societies with the IRI for for disaster managers for their humanitarian aid managers in the Red Cross So normally if if disaster strikes somewhere big big floods, then they They they react So they mobilize resources to to help people who are suffering for dessert from a disaster But they realize that if if they could be better prepared That would that would help them a lot because they always have to just scramble when when the disaster strikes So they in in talking with the people at the IRI Simon Mason included they they came up with this ready-set-go Idea where you could already get ready using a seasonal forecast you had some Some idea from a seasonal forecast that the season is likely to be above normal you could anticipate there may be a Like higher likelihood of flooding happening at some point in the season somewhere But you couldn't know where you couldn't have you couldn't be very confident about that But what you could do is Already things like you know train volunteers and sensitize the community or maybe enable an early warning system If you had a heads up already from the seasonal forecast Then this is the set stage here mid-range forecast This is where a sub seasonal forecast could come in that would have more specificity shorter lead time could Tell you more about the likelihood about of where A high-impact weather event a flooding event could could occur And if you see if you see that in the forecast and you can do things like well alert the volunteers that Really and warn the community that this is actually Quite likely and then go if you see the the tropical cyclone or the cyclone bearing down in in the weather forecast Then you can activate the volunteers and distribute instructions communicate to the the community to evacuate or so So this this was actually used in in the formulation of the the S2S project and in the the when we wrote the implementation plan this was a In terms of applications, this was one thing that was was was made use of to point out where it was that that S2S could Be of value in filling filling in the middle here between what you could do with a with a with a vague seasonal forecast What you can do With a much more precise weather forecast, but you have much less time to act here So as you go the other way you have more time to more time to act and this is a similar schematic from from Erin Coughlin of the The Red Cross Red Crescent climate center and it's just giving this idea as you go down From climate change to seasonal to weather forecast you get more the information becomes more specific But you have less time to act you have more time to reduce the risk With longer lead forecast so bringing those two those two things things together is a Lot a lot to do in developing These kind of applications So credibility can I trust the forecast? So one thing we always emphasize is that in order in order to make use of a forecast a climate forecast If you want someone to use it you have to make sure that that forecast is Is a well calibrated forecast and that conveys the uncertainty and that's where you really? That's where we really need probabilistic forecasts in being able to convey that uncertainty and that the user needs to know about that so whereas a Do a user might like to just have a deterministic forecast That that tells it gives them a best guess your your best guess of what the season is likely to be Like like like likely to hold in store what if you're really serious about using a Forecast then then you want to know well how like how likely is it that what you're telling me is is is Is is going to transpire so if you look at the IRI forecast maps you'll see for the seasonal forecast Many many white areas where there's just no forecast at all And that's because this is a calibrated product where we're only issuing forecasts in regions where those forecasts have skill and then where we issue a Forecast these probabilities should be That the forecast is well calibrated so that and this is something we'll talk about this afternoon But that the the probability of that that happened should be correct on average It shouldn't it shouldn't be it shouldn't be overconfident and biased so that you're giving very high probabilities when if you look at all the times when you you you issued a 70% chance of being below normal if that didn't happen 70% of the time in practice then then you're overconfident and Understandability that's that's an issue here with these forecasts and it's something we've we've struggled with at the IRI for for many years This is this is our sort of iconic product and we haven't really changed this over the years I mean people have complained about this a lot that it's hard hard to understand But we haven't really come up with a better way better way of of showing this information and this gets down to this this issue of How to make it how to make the forecast Understandable, and I thought I just flash up this you may remember this Schematic from Adrian's lecture. I guess probably last Monday. I probably so much has happened in between. I don't know but That must seem a long time ago last Monday when you first arrived So should the forecast be probabilistic or deterministic? You know someone from the IMD. I remember saying to me Well, the users it's what the users want that we give deterministic forecast because that is what the users want so Often that often that's the case, but this is this is the issue I That that was was shown by Adrian last week that it did what actually happens can be way outside of What what the forecast model is saying and there's no information about uncertainty conveyed in that So one way one way for what what how can we make that that information more tailored though? We think that using thresholds specific thresholds may be a way I said we've been sort of struggling it's struggling with this and one that I'd like to just show you now Quickly because I know we're sort of already getting into the coffee break now. I shouldn't oh no not coffee break The next lecture so I don't want to get too too far behind here. So Yeah Okay, well, I'll go I'll go for another another 10 minutes or so We thought about well, how could we Express this in a different way that would be on the one hand more understandable and on the other hand more salient that it would be Something more specific to what a user Could actually use rather than having a heads up below between What's the probability of it being below normal versus near normal versus above normal? That that's somehow for a lot of users that that's well Don't really know how to interpret that and so When we when we discuss for how should you be able to interpret a forecast like that we always show Well, you need to to look at your own data to see what what is below normal or near normal or above normal mean But I say this is over 30 years just ranking the data for your particular place and so if you have a 70% chance like in our forecast now for for For much of Indonesia that that would mean if your place in Indonesia has The 33rd percent are about 45 millimeters of rain Then that would mean that you have a 75% chance of it being less than 45 millimeters But could we could we generalize that so that the user can can pick their own Quantile so that's what we we we call our flexible format of identifying a threshold So if there's some threshold that users really care about like how much rain do I need for my rice crop or What's the what's the probability that this coming season? There's going to be that one year in five drought where my index insurance is going to going to pay out that they could choose a Particular threshold of rain for a particular amount and find out. Well, what's the probability of not getting that amount of rain? So rather than than this map that the corresponding map would be this one and So we call it flexible format probabilistic forecast and there's a map room in the iri data library for this Where you can go in and you can choose your your target time here is December 15th of February 2016 and here this is where you choose your your threshold and You can choose exceeding or non exceeding of a percentile like like the median or it could be the 20th percentile Or you can actually choose a an amount in in millimeters So if we zoom in now to you can just drag and zoom on that map and I've zoomed into Indonesia here And what I've chosen is what's the probability of not exceeding the 20th percentile? Which would be the the one in five Year drought and that that is what gets shown on this map So you can see that here in parts of parts of the maritime continent you have a 70% chance that you're going to be you're going to get that You're that that drought year So we think that this is something that could be More more salient and be able to target a particular threshold of User relevance as well as being something that people could picture more easily Think about what what's the chance of it being that one in five year drought or? Something like that. So we're hoping we're trying to promote This as an alternative way of querying our seasonal forecasts, but still if you I think if you go to our web statistics you'll find that We have a lot of hits on this map, but this product is is buried down somewhere and Not not many people not many people know about it and obviously it takes some it takes some explanation to be able to be able to use this and then what you can also do is you can click on a point on this map and You can get the PDF or the cumulative Distribution function for the forecast In in your region. So at your point. So for example, I could see well What's the probability of getting 300 millimetres of rain? And at this point that I chose Over the Philippines There's only a there's only about a 20% probability of that versus in in a normal in A normal year. It would be an 80% chance that you would get that so by using this tool Also, there's a number every year because if the forecast doesn't have information, then you would just be giving the chronological probability So that sort of gets around these white areas in the map where you can't say anything from our normal normal Forecast here you will be able to say you'd issue the issue the forecast in terms of what's the The likelihood from these historical data that you would get a certain amount of rain. So there's always information so More on credibility and understandability Getting back to the schematic. So proper dissemination of the forecast and communication with their meaning is also critical as I mentioned earlier And what I want to emphasize is this really co-production of forecasts with intermediary agencies This is something that we've been pursuing quite a bit at the IRI over the years With the IFRC with the World Food Program or with the national national Met Services. These are So crucial intermediaries Really the the legitimate intermediaries when it comes down to to country-level information or regional WMO regional climate centers like like ACMAID agricultural extension services Agricultural universities I mentioned the one in Indonesia and then training courses like this one or once once we've done with the WMO or the IRI to bring people from from the the agricultural or Hydrologic community together with with forecasters to to co-produce the information So I realize that well, I think you know, I'm a climate person who's not from the application side And someone who does application they should really be someone from the application side, but we think about The use of climate forecast for application, but really it takes both You have to get people in the same room from the from the user or intermediary communities together with the forecasters Only then can you share the right kind of information Sometimes it was popular to say okay. Well, we should have demand-driven information for to inform development of Forecast for applications, but then if you go and ask users To demand the information that they would like they would say well, I want to have a perfect forecast Of what is going to happen every day? So it can't only be demand-driven it has to be this this it has to be done in partnership And I think that if this there's one thing we've learned at the IRI It's it's it's been that that you have to work together In an interdisciplinary way to do this applications So here's a little schematic that came from an S2S workshop we had in in Korea that we had a discussion discussion session on application and Now we put in the middle here this intermediate users box here with some of these Agencies or groups that I mentioned could also be WMO commissions for for agriculture and hydrology other ones there national government agencies ministries of agriculture, etc or journalists and In terms of dissemination this was this was the product we actually came up with with the International Federation for Red Cross and Red Crescent and it was a mat room which allowed a one-stop shop for those humanitarian aid managers to be able to zoom into their region and look at Climate information from various sources such as six-day forecasts is from from from Ensep three-month forecasts from from IRI looking at Past conditions what's the changes in rainfall typical of Evel Nino for example or or recent climate climate trends in the region So it would allow it if this is an example of a dissemination Mechanism where also if you look at the way that the information The terms used here. These are also terms that were were were Were conceived with the with the users together so in this case There you'll notice that there's no numbers here or probabilities Heavy rainfall very heavy rainfall extremely heavy rainfall So we think that these these this mat room concept is a way if these are if these mat rooms are built In in partnership with intermediaries and users. It's a way to to help disseminate information and then lastly legitimacy This is something that's really emphasized by by the WMO that Forecasts they need to come from a trusted source that that one's obvious But that the national met services in the regional climate centers that they are they the legit legitimate conduit for for Forecast forecast for for national national usage and the WMO lead center For long-range weather forecasting multimodal ensembles is the the intermediary between the global producing centers And the national met services. So this is something that may have been meant. I'm sure it was mentioned last week that Frederick mentioned this that S2s databases lag three weeks in real time so that you can't use the forecast but that the the WMO lead center will Is setting up a prototype now or ready to to be able to distribute to distribute those forecasts in real time From several of the GPCs through the national met services So they they have a mechanism like that for the seasonal forecast and they're now doing it for for the sub seasonal ones as well So they actually access they could they actually access the same database at ECMWF And I and pull that pull that data without the three-week delay And so that they have access to the forecast in real time and later on that that that will be made available to the national met services So just as the this is my last slide before the summary And this is just the outlook here and basically the challenge is can can the S2s Climate forecast help farmers and others help avoid harm and disaster or take advantages of good years So this is sort of a societal outcome. It doesn't have to be production So you've seen PDFs like this for rainfall and climate, but here we're showing it for an outcome So there can be good years in the green there can be bad years in yellow and there can be disasters in brown in this this left tail here and can we can we use this the Forecasting on on seamless forecast if you like on sub seasonal to seasonal scales to help Avoid this kind of situation disaster and harm at the same time if if conditions are good, you may be able to choose higher use more fertilizer apply more fertilizer or better higher yielding seeds to to to To to get an opportunity from from better conditions So summary of the main points I hope to convey that the applications is really hugely massive multifaceted area and that there's much to be done in in I'm much new that Many new opportunities for better specificity and timeliness For for other decisions that people couldn't use a seasonal forecast for or people can't use weather forecast for that many other decisions come into come into play here and Developing seamless forecast across The the information on these different timescales that is really our challenge for the future much hinges on effective communication of What forecast can and can't say Include this includes doing training events like this and That proper calibration and verification become critical for people to act on the forecast and I'll talk more about this this afternoon what The challenge is what we want to do on thinking to do on the the sub seasonal forecast for for verification and issues of calibration, so I'll stop I'll stop there and Take any questions. Thank you