I work at the European Centre for Medium Range weather forecasts, but I work in the Copernicus department for the Copernicus climate change service, and indeed I'll be talking about

something called the European state of the climate, from data to information and back, and I'll mainly use this European state of the climate, which is a report, as kind of a way to illustrate where we're trying to head in terms of the services provided within C3S. So just a brief outline of my talks, I will talk very very briefly about the Copernicus

climate change service in general and the Copernicus programme as a little introduction, and then I'll move on to what in Copernicus 2.0, which started a month and a half ago,

we're naming climate intelligence activities, which mainly is concerned with climate monitoring, and this is kind of the work I'm most concerned with within C3S. But then I want to spend quite some time talking about how to kind of solidify the connection between these activities and the rest of the C3S service chain, all kind of in there,

under the big umbrella of APIN data, of traceability, of APIN access to code, and so on and so forth. But I think this will all become clearer during the presentation. So I assume that some of you at least are familiar with the Copernicus programme,

and for those who are not, the Copernicus programme is a really big European Commission funded programme. One part of the programme is a satellite programme, so taking satellites, so the sentinels, into operational production and sustained operations

forward in time. So that's kind of a huge data production part of the programme, let's say, but the programme also contains these different service elements, which are there to facilitate

the uptake of that data, but also to facilitate the uptake of data within their own realms. I think one of the key words that you might know if you know about Copernicus is APIN and free, so the aim is to have as much of this data as APIN as free as possible.

And for the satellites, this is the case, but for instance, the climate change service, we also of course have to rely on other data sources, and sometimes these are breakred and might come with kind of their own licenses, etc. So as I said, I work for the climate change service, and really our role is to

provide data and data processing to inform policy making and inform decision making in the realm of climate change. So using that climate data to understand climate variability now and climate change as we go forward and climate variability in the climate that has

changed already. So if we look at kind of C3S as a construct, I would say it's quite constructed around an infrastructure, and I will come back to that later, and where on the one end we have

data and data provision, so some of this comes from satellites, but some of it also comes from other sources. One big part is on re-analysis, for instance, which is a lot of the work I'm working on kind of took a start in the re-analysis, whereas other parts of the

program work more heavily, for instance, with model-based estimates, like climate projections or climate predictions. My work is generally kind of at the other end of this

long chain, so after a lot of processing and understanding of the data and what can you do with it, and the people in my group and in my team, we work mainly on climate monitoring, and we have kind of two key products, I would say, that we regularly produce, and one is the

monthly summaries, so these are information on a monthly time scale, and then the European state of the climate, which we publish annually. So I will just talk a little bit about these two products, and then I will come back to how I think or how we're trying to kind of strengthen the connection between these products and the whole chain that leads to them.

So we have this monthly climate bulletin, which is mainly based on data from re-analysis and from in situ, and we publish this around the fifth to the eighth of the month, depending on when

the weekends happen, and it's kind of along the lines of what many weather services or national metrological services will do on a national basis. They will look at the past month and put that month into the perspective of the longer term climate, so

was it warmer than average, was it cooler than average, were there any particular events that kind of stood out from the norm, or you know, how was that month in the kind of light viewpoint of climatology. So we have three, well actually four parts of this bulletin,

one is about surface temperature, one is about sea ice, one is about hydrological variables, and then one is based on an in situ monitoring product, which comes out a little bit later in the month, and now during Copernicus 2 we're trying to kind of bring these a bit more closely together than we had managed to do during the first phase.

And these can all be found on the website if you're interested, and what I kind of wanted to point out here is not so much maybe the content, so we talk a lot about global temperatures and we focus on the pan-European domain, and also possibly sub-regions within that

domain, but always on a larger than national level, because that's kind of where our remit lies. And within this bulletin, which is purely web-based, there's also kind of access to various types of information, so for instance the reference period we refer to, and also kind

of from a kind of communications point of view, the ability to download the images, but kind of more to the point of my talk today is the kind of access to data, and at the moment access to data for time series is done purely via a CSV on the website, whereas the data, so the gridded data behind the maps you can find in our climate data store, which I will come to in a second.

If we then look at the annual product, this is much much larger in scope, and we also make use of kind of the whole range of data sources, which are provided as part of the C3S program,

so this is based on satellites, and also on in situ, again on re-analysis, and also to some extent on model-based estimate, depending on what variables we're looking at, because as we all know some things are well observed, some are less well observed, and in order to kind of make central statements about certain events, you need to kind of re-resource to all of these different

types of information, and this report is kind of with contributions from institutes across Europe, part of the Copernicus kind of larger kind of program, and we report on the past year

in the climate context, and we for instance look at very regular reporting on temperature and precipitation, and so on, so there's kind of a continuity from year to year, but then we also have more topically based parts of this report, which will look at different events for instance,

or kind of looking at how one particular event might have affected, or be represented by different parts of the climate system, and so the last one was published in April of this year,

covering 2020, and I just want to give one example of the type of topic that we cover in there, say one example for 2020 was the very warm winter at the start of the year, so going out from 2019 to 2020, and it was quite by far the warmest winter on record, and especially this

warmth was concentrated to the east of Europe, and then in the kind of monthly bulletin we look at very, I would say quite fairly simple straightforward variables, whereas in the state of the climate we try to go more towards indices which might have like a further meaning than just kind of mean temperature let's say, so here's the example

of ice days, which was extremely low during that winter in that particular region, so an ice day is a day where the maximum temperature of that day is below zero, but then we also look at how that event had other effects, say for instance on sea ice in the Baltic Sea region,

and also other things, and as it pays to maybe the climate bulletin, we do try to kind of give a traceability back to the source, so in this case the data behind all of this, and so the approach we have within this report is to have one quite general summary

of the content of the whole report, and then these individual sections, here again the example of the warm winter and it's all web-based, and then you can kind of go step by step deeper and deeper into the report, see which data sets were used for the different

sections, and then get to these data sets in this infamous climate data store that we'll come to in a second, but by doing that we're kind of covering the two ends of the spectrum, so we're looking at the report on one end, and we kind of give indications of this is the

data we used, but not necessarily of all the kind of processing in between, and what we like to strengthen is for that chain to become more transparent to the user, so if you see something in this report you may want to try to reproduce it, or you might want to repeat the same thing for another year, or whatever you may want to do, you may want to know what processing was done

rather than just knowing what data was behind it, so that's what I'm going to talk about in the remainder of the talk, is how we're trying, you know some of the building blocks that we're working on to make that happen, and this is quite complex if you want to try to do it

everything in a report like this, but where I kind of in some places in the starting blocks and in some places more advanced, and I will just kind of give some examples, so I've mentioned it a couple of times, the climate data store, this is really the backbone I would say of C3S and its services, it's a data store but it's also a processing

facility, mainly for kind of processing at a reduced level, so it's not for running climate models for instance, and at the moment we have almost 100,000 registered users

and there's about 70 terabytes being downloaded every day from the climate data store and if you were to visit kind of the welcoming page of the climate data store, you will be met by this, and I will kind of deep quickly dive into some of these areas

in the CDS and how we can strengthen kind of this service chain and connect it better to the report really. So one first step, I mean this is quite an obvious

one, which is implemented already, it's of course the data download, and I didn't quite dare to take you into a live session of the CDS because you never know how these things work online, so I've decided to do some short movies and I can just walk you through

where you would end up if you were to click on this link. So everything in the CDS is kind of structured in the same way, you'll have like an overview for the data set, you'll find documentation, and then for the data sets obviously you'll find download data widgets, which you can either do in this GUI, so you can scroll down and make your selection of

different things, and in this particular case I've chosen the data set which is behind the monthly climate bulletin that we have, and as I said kind of earlier at the beginning of Copernicus you always have to kind of accept the license agreement for this data,

and in the majority of cases we try to keep the Copernicus license which is a very open and free license, but for some data sets because they come from other providers and this is not always the case. So you can download the data by clicking on this submit form,

and it may take some time, in this particular case I've downloaded the data earlier, but you can also go back to your old form and what you can do if you're not interested in this GUI interface is to show the API request, and then you can download the data via API and instead of doing one month at a time you can do a couple of more months at a time,

and there will be a limit, kind of an advice limit on how much to download in each kind of request. So you have kind of the data download on one hand of the kind of raw original data set,

and then another way to connect with the report is kind of a visual one, so within the CDS there are also a number of applications, some of them are very complex and some of them are relatively simple, and this example here is called the climate bulletin explorer which basically just allows you to explore the data set which is using the climate

bulletin and kind of look at the same variables as in the bulletin, but you can go back in time or you can look at kind of various regions. So if you were to go to this application, you can kind of choose the different things that also appear in the bulletin, but then

there as I said we look at the kind of pan-European level and you can mainly kind of see we might look at these kind of sub-regions of Europe, and if you zoom in on this application and you can kind of choose one of these sub-regions, for instance Southwestern Europe,

and kind of get the same graphics as we show in the bulletin, but for other regions because we don't want to overload everyone with 1000 graphics everyone, so there's a pre-selection which is kind of made, and then finally what this application also allows you to do

is to explore the data set kind of further, and for instance looking at nuts levels across Europe, but here I would also like to say that here you can look at the national scale, but we may not have compared this data set against like reference temperature

data sets for instance for the different nations, so it's not something we report on in our kind of monitoring, it's just one way for people to explore the data further. So we've gone from downloading data to viewing the data, another step of course is processing the data and getting closer to what is actually shown

in the report, or for instance this case with the time series within the viewer, you can't actually download anything at the moment, and I just want to give one example which I think is quite nice because it's quite simple, so it kind of illustrates the concept quite well,

and that's an application which calculates daily statistics based on a data set that has hourly statistics, and it's a huge data set, so from a kind of download point of view if you're not interested in the hourly data you would want to process it before actually trying to start

a download, and if we visit this page you get very similar kind of thing as you would have done in the two different examples with the overview and the documentation, and then instead of a data download for the application of course you have the application button, and in this particular case it's looking at a couple of different data sets from which you

can then extract daily statistics for months going back in time, and I have to say I ran this before so it's not as quick as that, if you would just run it normally on your own,

but then you can download the data here, you know file by file, and you can also, if you click on the source code, see the source code behind this application, and if you wanted to you could replicate the application within your own environment and change it, but what you can also do, which I will show you in a second,

is that instead of downloading this manually hand by hand, you can also use an API script which will basically replicate this so you can then loop over all the years and all the

and all the months etc, and the reason I'm showing this is not necessary to tell people you need to take a look at this particular application, but just to kind of illustrate the kind of different building blocks we'd like to see in place in order for the whole chain to be traceable and reproducible by a larger audience. So this application is online,

still the one I showed before, an outdated version I would say is online at the moment, and a new one will be released relatively soon. And then finally, in terms of

these kind of building blocks, within the state of the climate in particular, we look at climate indices quite a lot, so for instance this example of the ice days, and then of course you can go one step further and not look at purely on indices purely from a climatological

point of view, but maybe indices that kind of become more relevant for specific sectors, and there's a lot of different examples of this in the climate data store already at the moment, and one example which is currently in progress is looking at heating and cooling degree days, so it's basically an index for understanding for when heating is needed of houses or when

cooling might be needed, and this is I think one of the first applications that we have where we really try to make the whole workflow from the data until the process index traceable within the toolbox and the source code itself, whereas in Copernicus 1 there was

a lot of focus on creating these indices and putting them in the CDS for people to download, and now we're trying to move towards a more workflow-based approach where you could

potentially redo the same calculation but using different data, so the example here for instance for the heating and cooling degree days is that they also use climate models to go forward in time, and of course with a relative regularity there are new climate projections coming out,

and the idea would be that you could then kind of pull those in and rerun the same workflow again based on the data that's in the CDS, and then the idea here when we come coming back to the state of the climate as well is to be able to kind of harness

some of these statistics for the reporting that we do in the state of the climate, and so this is kind of really looking forward in the kind of work we want to do in the coming months and years. So I've kind of gone through these blocks of the service chain trying to show

what parts are maybe there in the CDS already, and what parts we're kind of working on, or some of the ideas of what we're trying to make kind of transparent and accessible, and as a final thing I just wanted to show you the climate adapt portal by the EEA where we've

been working together with the EEA to kind of transport some of these indices not into a report but into a live environment, but in a live environment which is not as complicated as the CDS and as such maybe more useful or digestible by the EEA's, the European Environment Agencies,

customers and clients and the people they work with, and of course a portal like this one will then benefit a lot from having these different workflows more adaptable and

updatable going forward. So with that in a way I'd like to conclude already. So as I said there's quite a lot of ongoing work in different parts of the programme,

but really working towards, well from my point of view and this kind of climate intelligence activities of strengthening the link between the reporting and the kind of data behind to make it more traceable and transparent both in terms of data sets and applications, but then also

vice versa to make use of more of the kind of mature offering let's say of other C3S activities so that we can harness more things which is more maybe a sector relevant and more targeted than we've been doing in the state of the climate in the past.

So that's kind of like an outlook for that side of C3S and for those who might be familiar with the CDS and have worked with it already, and I also want to say there's also a lot of work going on in the CDS itself for improved user experience and performance etc,

but it's not really my place of expertise, I can't really at least not make a presentation about it and go into the details of that, but I just want to say there's a lot of work ongoing and it's all very exciting with Copernicus 2 just starting and knowing that

there's another seven years of program ahead of us where all of these things can be implemented and updated as we go along. And with that I want to thank you for your attention and just

leave you with some links which links to the climate monitoring climate intelligence activities and I'm very happy to take your questions or go back to any of the slides if you're interested. Thank you very much. I hope you heard the audience.

No, I didn't, but that's fine. I'm so sorry, I had my microphone muted so I don't interrupt.

Yeah, there was a clap. Very good. So we are now, we have quite some time for questions. Let me see, they will pop up. Let me start with the question. And the, when you make some modeling some climatic variable, like we do a different,

we model here different environmental variables and like modeling soils and vegetation and land cover. We tend to provide also measure of uncertainty per pixel. But that's something I rarely see actually with the, with the climate data. Can you say something about that?

Like if somebody would like to know, well, I have a pixel now and it says some temperature, number of days under ice and what's the uncertainty of that? Yeah, so I think it depends a little bit on on what you, the level of where you're at with the data. So, if we look at the data,

within the climate data store, for instance, there are quite large, what's it called, like documentation around, around the data sets and their associated uncertainty. And some of them might also come with uncertainty variables, whether it's

an ensemble of realizations. So for instance, one of the in-situ data sets, EOPS has a kind of, it's been run several times. The way they put that gridded data set together. And because of that, you have a kind of uncertainty estimate around that.

And then something I didn't go into too much, but there's also quite a large EQC function. So in C3S, which is kind of rolling out, I would say at the moment. So if you go to some of the data sets on the CDS, and it won't be all of them at the moment, I can probably show

if I just go back to the example I had here of the data set, you see that there's a thing that says quality assessment. And so they won't be necessarily like an additional data set that

you can download, which gives you the point grid point based uncertainty estimate. But there will be documentation that allows you to understand that uncertainty to some extent. And when it comes to the state of the climate itself, what we try to do as much as possible is to use more than one data set. So to use a multi-source assessment and to gauge

the uncertainty that way as well. Okay, next question. So I really like this combination of, you know, your data, your really proper user-oriented data service. So you make it easy

for people to grab the code and extend it and really speed up download. But what about cloud solutions for data? What about cloud alternative and stack? Yeah, so some of these words mean nothing to me, but I can answer partly.

So first of all, the CDS is a cloud in itself. And there's also a way to connect it with some, well, with one of the diocese within the Copernicus program. And then of course, at Easeum.wf we're also working with the European weather cloud. So there is quite a lot of work ongoing kind of in the cloud side of things. But I can't really go beyond more into

the technical details of it, because it's just simply not my expertise. But let's say GDAL, I mean, you're familiar with GDAL. So, you know, how it's easy to load this data using GDAL.

I don't know. The maps, like, you know, so the NC extension, so it's an FCDF. So I will assume because it's a map, it should be GDAL compatible. It should be relatively easy to load the data in QGIS, I don't know. I think, I mean, I think yes, if you load it, but the question is,

where do you dock in the cloud? So for instance, in the CDS itself, you can't access that cloud, say to say, without downloading the data. So your processing would have to be done like within the kind of infrastructure of the CDS. Whereas if you were to access in theory via the diocese,

but I've never, I've never worked with them. So I can't say how well it's working, then you would maybe be able to work more directly with the data. Okay, the next question. So you can do modeling locally, like in a country, the country based

meteorological and climate services. Then you have the European, that's basically your, your, your institute, and then you have global. How do you harmonize the tree? And who does that? Well, so it's a good question. There's kind of a lot of harmonization where we try to do

via the auspices of the WMOs, the World Meteorological Organization. And so when it comes, for instance, to the state of the climate, there is a global statement that comes out every year and kind of different people involved in that. And I've been, for instance, I've been involved in the past few years. So we try to align

that and make sure that the kind of communication on key things that are communicated at all levels kind of align. And then when it comes to the national level is kind of similar, we try to work with WMA regional center. But I have to say is also is quite,

it's quite complex to make all of these things align. So I think the important thing is to kind of try to use the same methodologies, say that kind of automatically things are quite aligned. We have a question from the audience. Please maybe say the question and I will ask. Yeah, I'm looking here. I'd like to apply it.

So the question is, for example, you have for global data, and we are now pleased

from the audience is needs data for us, let's say, but there's a US also, US has some data there, I don't know, NOAA or something, they made a data set. So how to make a decision which data set to use? I think that really depends on the application and what you try to do.

So I think if you are, let's say it's the same variable, you know, like, I don't know, temperatures, rainfall, let's say it's the same climate variables. And so what would you maybe maybe we have to do some ensemble estimate or what do you recommend

usually in these situations? Well, so I think it does actually depend on what you try to do. But like, I generally, I think if you use more than one source of information, then you have a stronger case than just using one data set. But then I think if we take the example of the US, if you're

interested in understanding whether NOAA will say that there's been record warms in contiguous USA, they will make that statement based on a reference network. And so you will get one number out. Whereas if you would use, let's say, the reanalysis, which covers the US, you will not

get the same number because you're not measuring exactly the same thing. But it kind of depends on what you're interested in. If you're interested in the trend, probably the best thing is to compare both of the data sets to see if there are differences or if they agree. And if they agree, then you can be more certain of your conclusion.

Thank you. A question about resolution, spatial resolution. So many climatic variables like rainfall, you know, you wouldn't go in, you don't need like a, I don't know how to meet the resolution because it's like a, it's a feature like atmospheric. So, but

some some for example temperature now there's a new European Space Agency has a new project. I think it's next year. I think it's called LST or something the, the temperature satellite, and it will be I think 50 meter, 50 meter temperature measurements, 50 meter resolution,

how much do you think it's needed spatial resolution for some climatic variables, how much is it needed? Do you think it's most of variable sexually or fine with the core resolution or? Again, I mean, this is this is a really, this is a really boring answer because I think again,

it really depends on the application you're trying to use it for. So if you look at LST, for instance, is the land surface temperature, say, actually, personally, I think quite often, there's some confusion in the communication, because things will be treated about land surface temperature, and people get very excited about very high temperatures.

Not realizing that people usually report on when it comes to temperatures is surface air temperature, which is generally quite different. And I think the resolution depends on, you know, if you want to understand the trend of change in a, in a city or country, then

maybe doesn't really matter. You don't need exact location, but if you need to do something like precision agriculture, based on based on that, then, then the resolution becomes important. Let's say if you want to track like fire danger, or, you know, something can like happen

locally, you know, so then then possibly spatial resolution will make a huge difference. Just the last question about IBM, I think they have a system called graph or something. So high resolution weather forecast system. So how is it? Do they really get a much better

accuracy in forecasting? Are you? Do you? Are you people jealous for the equipment they have and the system they have? What is your impression, if I may ask? You have to ask my colleagues in the forecasting department. Okay. Okay. But they, they kind of have, I think, a global system

at one kilometer. And the last thing I read about it. So, and they claim that, you know, they, by doing a more higher spatial resolution that they also increase the accuracy. Yes, I'm not seeing the in every month or so there's a score comparison between the different

centers and the different models out there. But I've not seen any of them for a long time. So, okay, there's a question for my forecasting colleagues. Okay. With this thing, I would I need to we need to stop because we have another speaker

coming. I would just want to thank you for your time and for connecting and hopefully maybe join our next workshop, which is in June next year in Prague. Yeah, it sounds great. Again, on the European data cube, environmental data cubes. So thank you one more time for your talk and good luck with your work.