FOSS4G 2021 - The ESA-EC Open Science dashboard Rapid Action on Covid-19 and EO

your dashboard corporation previously she served as a project official with the european union satellite center developing solutions for geo-intelligence and also has a phd in star image information mining and stochastic analysis so anka if you're ready you can share your screen um can you see my screen already um not yet okay okay [Music] oh yeah i think okay now it appeared yeah okay great so thanks so much can i start oh yes please you can start okay thanks so much nicolina for the introduction uh and thanks for for being here to to listen to this talk so um i will present today the project called rapid action on covet 19 and earth observation i just like to say that um this is a collaborative project so i'd like to acknowledge my colleagues first patrick griffiths and stephen michael who you see also on this slide but also the large number of contributors from european industry european commission copernicus services and so forth so just a brief um introduction to the project um this project was started last year um on the onset of the pandemic and it's a joint initiative of isa so the european space agency and the european commission and the scope was to act pretty fast to try to see whether earth observations could be used to measure or understand at least the impact of the pandemic on different aspects of the of the society economy agriculture and so forth and to communicate these findings to the general public and potentially also decision makers so we had four main focus areas climate environment economy and agriculture to start with and the objective was to try to launch in a relatively short time a public platform using open source technologies um based on data that comes mostly from the copernicus sentinels as well as third-party missions through the earthnet program using of course the platforms that isa has been developing in the late in the last years and machine learning and ai to extract information from us observation data so this project was launched um in early june 2020 and has been ongoing since um it's available at race.esa.int and you have the github url as well on on the screen and currently it has reached people in over 127 countries and that was quite um good news for us so the main question that was let's say our starting point was how to get from the large amount of earth observation data that we had available together with other sources of information transform them and extract let's say relevant actionable information that can be easy to interpret easy to understand by the general public and how to do this very fast so we aimed to build a product that could give let's say a first impression view of what is happening due to covet with our society luckily we had um already available a number of activities um within isa within our department of um data applications with a number of industrial actors so this is largely a collaborative project and through these um different actions that we had already ongoing we were able to let's say distribute workload and produce this platform that you should be seeing on the screen basically what this platform does is a web application that has this central map and a number of filters that you could see before and you can browse by country and by a number of economic agriculture and environmental indicators all of the information being derived from earth's observation data you can scroll through the locations [Music] inspect the various indicators for example this one that is shown on the screen right now is actually data that is not coming from mars observation it's um it comes from google it's mobility data so integrate a lot of different different data sources um and try to put everything within a context we have also three community contributions so this is an open source project and we've launched number of competitions since last year and continuing also this year competitions where participants can reuse the resources that we ourselves are using to develop the project and try to propose some additional information that could be added to the to this dashboard so one of the key assets that enabled the development of this project was the eurodata cube so the eurodata cube is what we call the eo information factory and it's in essence platform technology um so it's um the euro data cube itself sits on top of cloud infrastructure so it's using uh the dss but also aws and it's interfacing with a lot of different um other platforms that are providing different data sources different types of data so we have in c2 data um as well as openstreetmap but also a lot of a lot of various different providers on top of that we have a number of services that enable not only the data access but also processing um as well on the fly uh we have vector data service we have the possibility to um develop and run applications and finally um the whole system is interoperable and exposes the information via apis uses ogc standards so um the information can be easily transferred from one system to the other so the main data sources that we used um first of all uh we had the copernicus services the copernicus sentinels um so basically the indicators that we show on the platform are derived from sentinel one sentinel two central 3 and sentinel 5p but also a number of indicators rely on higher resolution data that comes mostly from third-party missions such as isi for example or play ad we have integrated as well contributions from a number of kubernetes services um atmosphere uh monitoring service and the marine service and as well a number of other open data sources including statistical data health data mobility and so forth we do have some indicators that have let's say underlying non-open data such as ais or mobile analytics but the majority of the data is open so just to give a few examples on what kind of indicators we show one of the most known environmental indicators that we show relates to air quality so this is an indicator that was produced by a number of actors that you see here on on the screen you can see their logos here as well as and let's say complemented by data that comes from ecmwf so here we relied on data from sentinel 5b and the processing services from the eurodata cube and the statistical api to produce global maps and citywide air quality trends so global maps such as this one that you see on the screen [Applause] where users can inspect so they can browse through time and can inspect the differences in the concentrations of no tube as well as cos you always i'd add it later on and what i would like to point out is that besides these maps we also um so the users have also the possibility to generate their own time series using the statistical api from the euro data cube so this statistical api basically enables us to compute statistical analysis for a long time series and to provide based on a given geometry and a selection of statistical measures to compute that those statistical measures for the earth's observation data that is in the back um just to show you how this looks on the platform so on over the map view um the user can just draw an area of interest and then by pressing this little chart button here they activate the statistical api and the return of the api is basically the different statistical measures that are then plotted on the interface another example is on water quality so here the teams of cnr produced um chlorophyll concentration and total suspended suspended matter maps based on central three data as well as station data so maps uh sorry um apart from the from the maps will also include uh time series um such as this one where users can inspect interactively uh what is the weekly climatology here we refer to chlorophyll concentration and compare for the different dates and whether the water quality was higher or lower than the expected value um a next short example of um indicators that are using also machine learning um together with platform um technologies so this is an indicator that shows the variation in the number of park airplanes on sentinel 2 data the indicator is produced for around 40 airports distributed across europe and apart from the interactive charts um that show whether the number of park planes is higher or lower or normal compared to a given baseline the user can also understand whether this effect is due to covered or not so then back on the back of the of the charts we display the periods of lockdown uh that are taken from the open data provided by oxford on the string covered 19 stringency index and finally the users can inspect as well for each of the dates on a map of the detections that were retrieved by the machine learning algorithm it's based on sentinel 2 data so we're really reaching the limit of the detection with this with this indicator uh i will just present briefly two of the newest indicators that we've released so this is an agriculture indicator that was produced by colleagues in vista and contributors to the food security thematic exploitation platform so this is an indicator that is presented for 19 european countries and it shows them on the same charts the evolution of the harvesting as it would be expected based on the epsilon platform of vista compared to what was observed by using sentinel 1 backscatter and coherence information and this is done for winter cereals and winter rapeseed finally one of the latest releases refers to an indicator an economic indicator that presents crude oil storage across four european clusters so these clusters are the uk and ireland cluster the northern europe including belgium netherlands and western germany central eastern europe with croatia austria and northeastern italy and southern europe so this indicator is a composite indicator that uses central 1 and sentinel 2 data as well as additional root information and ais data so what we show on the platform are the different sites that are part of the of the different clusters and for each of these sites users can then retrieve graphs such as this one where we show in the continuous black line the monthly cluster storage average and then with the different colored dots then how that cl how that particular site um [Music] is also the status of that particular site with respect to the cluster so for this um chart that i've shown here that is in the northeast of italy we can see um during some stringency some um lockdown period in croatia that's the cluster the level of the oil in the cluster and that the specific site was much higher which means that the trade was not as expected just to show you two of the community contributions and to tell you that you can learn more about this if you participate in the ease of fee week that is taking place on the 5th 11th to the 15th of october so this is a community contribution that was selected as part of a contest last year it is a flying plane detection based on sentinel 2 that exploits this artifact this rainbow effect that is created by moving targets at a given altitude in the sentinel 2 imagery so this is an indicator um of um uh transportation so mobility uh and can be relevant for covid um by helping us understand how the disease spreads or if the containment measures were effective and so forth uh it it relies um so the the development of this of this indicator relies as well on a lot of open data not just from the sentinels so sentinel 2 but also open sky data that was used for validation finally we have a second community contribution that is exploiting the effect central 2 data this time for targets that are moving on the ground and that are um large enough to be detected such as trucks so we have this indicator developed for the whole um europe and i will just show you how it works and how again the use of the eurodata cube platform [Music] to generate time series on the fly so what we include here is the map where the user can draw an area of interest and can generate for that area of interest a chart it's a comparative chart showing the number of trucks that were detected for that area so on that road on that highway uh between 2017 and 2020 and then the individual detections can be also displayed or inspected so again you can learn more about this indicator by taking part in the few week there is a side event on the 15th of october starting at 11 am so you can register for free here finally um i would like to point out a collaborative feature that we've recently released uh it's called the custom dashboard so basically all the indicators that were that are present on that are displayed on the race dashboard can be combined and integrated as well with other information that is relevant for a particular user to create custom views so if you just i don't know if you can if i can switch screens here but you can maybe access later to these um urls that are shown on the screen so basically um this is a simple um from the main interface the user can just select which indicators um he or she wants to add to the custom view and then edit this custom view to include additional text or images or whatever descriptions one considers to be relevant and then publish this online and it's a collaborative feature so to users that have the same edit link and collaborate on it and create their own stories i'd like just to end by pointing out that we do have challenges open right now so the latest challenge was open on the 1st of september and it goes on until the 30th of november um it's a three-stage challenge so the idea is that first um there is an let's say the definition of the idea so indicators that can show um socio-economic impacts that can show um environmental impacts um as well as indicators that can help us understand where we stand with respect to the recovery so coming back to normality let's say and there is as well one theme that is completely open so um anyone can can propose indicators that that have not been yet proposed uh or bring a new idea so um in the first stage of the challenges uh it's just the idea definition then if this idea is pre-selected then it goes to an upscaling stage where the indicator has been let's say being refined and upscaled to the whole of europe using of course the classroom technologies that we make available and the third stage is integration so the ideas can get integrated like you've seen the the ones that i've shown before with the trucks and the flying airplanes they get integrated into the dashboard so the resources that we make available uh first i'd like to point out that for all the competitions that we've launched before we've created quite a large library of open tutorials uh jupiter notebooks uh where the users can replicate um most of the indicators that are being shown on the platform so they can exercise how to access the data through the your data cube how to launch the different processings use the different apis and produce those graphs that you just saw the pre-configured pre-configured workspace is um provided by by our technical team so i think this is uh it uh all i wanted to present so i'm looking forward to your questions thank you um thanks a lot for this really excellent presentation um i guess we have a lot of questions from the audience but i'll take the chance to ask the first one so you showed some really nice examples how you involve the community and basically use their knowledge to um for various purposes and my question is what challenges did you face in managing to involve this community and to really be engaged because i'm myself trying to involve different communities also indigenous people and so on and it's always a little bit of a challenge for me how to involve them and motivate them to to be willing to participate and contribute yeah thanks nikolina for this question it's actually um it was one of one of the pain points as well for us so we it was of course a learning curve and uh i have to say that we have um let's say um a sister uh collaboration with with nasa in jaxa the so-called eo dashboard it is basically a second instance of this dashboard but extended globally so not just for europe and there we collaborated a lot with with the space apps teams of nasa so they have a global reach we had a hackathon this summer that managed to to create to to bring together 500 teams so about 5 000 participants um for for the european challenges let's say so for for these challenges that we've launched now the first round of challenges that we launched were let's say quite quite specific and demanding so what we what we tried to do to to get more more participation from the public is to open up the themes a bit so not to make it so restrictive and to to provide as many resources as possible so now for these round of challenges we have created a detailed technical guide that shows the participants how they can access all the resources where they can find the data what this data is about how it was processed um how they can interact with the tool so documentation was one of the things that we we learned was useful and providing the participants as well with the opportunity to ask us questions so that's why also in the fee week in the side event that we we have on the 15th we invite those that have subscribed that have signed up for the challenges to come and to ask us questions uh try to validate their idea and so basically promoting this dialogue with with the experts yeah thank you um so i guess we can proceed with the questions from the audience awesome excuse me great presentation anka really interesting platform i'm excited to play around with it um when we finish this the phosphorchy um one of the questions from the audience is that this platform seems quite relevant beyond the scope of the pandemic are there plans to continue maintaining and developing it in the future thanks that's really a great question and yes we are actually in process of of evolving let's say this platform especially the trilateral cooperation with nasa and jaxa so where we have extended this international cooperation for uh for the next year as well and are looking to go beyond hopefully hope it will end sometime soon so um we're looking to address different more environmental and climate focused themes so yes it's ongoing that sounds excellent um really good news another question for you how is the indicator calculated is it through the dashboard or is it fetched from another um data source i guess kind of on the fly versus something stored yeah so some of the indicators are calculated on the fly so those time series that are computed from the maps for example for the air quality they're calculated on the fly based on the basically there is an averaging done on the area of interest that is defined by the user but other indicators are um calculated um let's say outside of the platform and they're just accessed uh by the platform so they're regularly updated great thank you um another question on the chat here um is there any information on the gendered impacts of covid19 available through the platform and through gender disaggregated data that's a great question uh we haven't addressed this topic yet um but um in the hackathon that i was talking about a bit before i saw the eo dashboard hackathon with with our nasa and jaxa colleagues we received a large number of different ideas and some of those were addressing not necessarily gender gender differentiated impact of coffee 19 but they were looking at um different [Music] let's say vulnerable populations so um looking at there was one team that submitted a proposal um an idea on um indigenous populations of the impact for of the coveted indigenous population excellent really interesting um those are all the questions that i see now um in the chat if there's any others from the listeners please feel free to pop those in i think we have time for maybe one more question um i i will admit that gender question is mine i'm curious i haven't had a chance to play around with the the platform yet but is there any kind of data sets in there that um do have either gender or um vulnerable population data sets like um the differences in male female mobility or indigenous groups or other minority groups [Applause] so there are several data sources that are not really open i'm thinking now the mobile analytics data so that one is as much as possible anonymized and was used mostly to to describe the impact on agricultural production or agriculture output so basically the information uh was or the relevant information was whether there was available workforce on the fields or not so without discriminating between the different genders uh the types of socio-economic data that we include are just population data so population density which is really aggregated so we can cannot really disentangle those um that at that level but um we do have uh so we do encourage contributions so if you have ideas of what kind of data could be relevant for this uh there is also a feedback button on the platform so you can send us our feedback directly from there and just submit a proposal of what data you you would find relevant and would be happy to have a look at that great thank you so much

Channel: FOSS4G

Views: 44

Keywords:

Id: IxvQbHknPls

Length: 28min 35sec (1715 seconds)

Published: Thu Nov 11 2021