 Okay, so now we have Timo de Guerre, a Brockmire who is talking about the emissions API. Yes, thank you very much. Yeah, thank you for letting me speak here at FOSSTEM and present our project Emissions API. To add a little suspense to the presentation, I changed the title a little bit, Emissions API, or How to Get Engaged in Public Interest Tech, because I wanted to take into account a little bit of our origin story, how we got to make this project because I found it interesting to share it with people. Afterwards, we also get a little bit into the details of the data and the technical aspects. But first, I want to show you what Public Interest Tech is about, because we first got in touch with Public Interest Tech when we heard about a call from the Prototype Fund in Germany, that is a federally funded program that lets a small prototype of a project they want to develop, and they proposed this framework of Public Interest Tech. It is basically a very broad and simple to understand concept that you develop technology that serves the public good. So it adds a little bit to this floss idea that software is not only open source and liberate, but it also has some kind of social meaning for the society. And we as developers as a small group of colleagues, we found that very interesting and we were immediately hooked by this idea because we always found that, well, it's nice that we know technical stuff because of our day jobs, but somehow we would like to also put it in use to solve problems that maybe have a meaning deeper than it goes further along than just the technical aspects, but more solve problems of society as well. And so we really like this idea and we said, well, we just want to apply to this call, to this funding call, and we're going to pitch an idea. And the first ideas we had was, well, have you heard about the Satellite Sentinel-5P that has emissions data? And we had heard of it and we thought, well, that's really cool. There's open data about emissions. And we built some cool tools with it where you can maybe analyze policy effects on emissions and maybe build some cool visualizations. And so the Satellite Sentinel-5, they provide a lot of data products for emissions. It's a program from the ISA, from the Copernicus program. And it's basically spectrophotometric measurements where they can analyze the wavelength and then they can calculate the concentrations of trace gases and emission data. So we were then, had some ideas what you could do with this. So we were thinking about maybe tracking ships at sea because there are no sensors and ships that use a lot of heavy fuel oils and there was supposedly a reduction in national ban on heavy fuel oils to reduce sulfur dioxide emissions. And we thought, well, maybe we can track that or think of maybe diesel bans in city centers and we could just verify how policies would influence actual emissions. Only that we very quickly realized that it's not that easy. So you can just take this data and analyze it easily and put it maybe in some visualization. So we thought, well, this is open but it's not really easy to access. So the good thing is that it's there. So this is really awesome. We can access this data. And we also don't have to do all this fancy analysis of the spectrum analysis. So the ISA already gives that to us. It's pre-processed data. And the problem that we realized is that it's packed in binary data sets that are not really too easy to employ in a program or in an interactive visualization, for example. So these are NC files. It's a scientific data format. And when you filter this data, you get chunks of files that represent one flyover of the whole satellite. So if you are interested in a region, you also have to do a lot of processing from the data that you already pre-filtered. It's pretty large files and also generally there's a lot of data processing involved beforehand. So we took a step back and that was actually when the emissions API was born. So when we came up with this idea, we need public infrastructure for open data. We need an easy access to this satellite-based emission data. And we wanted to build an infrastructure service that provides this data, that takes it from the ISA, does some pre-processing and gives it to a user who wants to use this data in a more easily to employ format. Also, we see ourselves a little bit as a data literacy project because satellite data is not... I'll come to that later, but there are a lot of peculiarities that maybe you would not expect when adjusting of a satellite. You think, well, there's a lot of data around with pretty dense and you can maybe make some nice visualizations, but there are also some constraints that are important to know. Yeah, so I would like to dive into this and to the more technical aspects of the data. How does this work? The satellite actually flies over the Earth and produces scan lines. So you can think of it maybe as a flatbed scanner for the Earth and you actually get a continuous picture of the Earth around the orbit of the satellite. And when you get one of these data sets, one of these files and you just plot the data, it's something like this. So here you can already see some general aspects of this data. First of all, we filter for Germany here, but you get a lot of Africa and Arctis as well and you get nothing, for example, of the North Pole. This is simply due to the fact that the measurements are based on light and there's no active light source on the satellite, so it needs to have sunlight to get data. So you will never get data at night, for example. Also, you can maybe see that there are some holes or that it's not as smooth as you would expect the line in general. Everywhere where there's clouds you don't have data and this is just some of the things. So now that when you think of how the satellite flies over the Earth, you don't have a measurement for each time of day in every place but you actually get this line that moves over Germany in this case here, for example, and you see where you have the measurements. So for Germany, it's a time window of maybe let's say 10 minutes where the satellite is passing Germany and you just get data for that time, that location. So when you want to, for example, measure emissions during rush hour and the satellite is not there during that time, you simply don't have data about that, so you cannot have comparisons over one day, for example. These are some of the important things to consider about the data. Now I'm going to switch to the technical side and our architecture. What we did, we downloaded the data from ESA. We pre-processed it in a way that we can use it more easily. We cache it in a way or we save it in a Postgres database and we will provide a service as a rest interface where you can query by region, by time, and you get a JSON in return that you can easily put into one of your usual JavaScript frameworks, for example, for visualization. I have to say that we are doing this since September and we are now in a state where it kind of works. You can check out our homepage and our UI. We have a working prototype. At the moment we only have the carbon monoxide data in our database because it is what we started with, because it was easy to get going with it and we still need to add some more of the other products. We have a lot of examples already on our homepage where you can see how to use the data, how maybe to make visualizations with this data that we have. Of course, we also needed to develop some tools around this service and one of them is Sentinel-5DL as a download library that is on PyPI where you can filter and download Sentinel-5 data automatically. It already works quite well. If you just want to think around yourself, you can also use some of our libraries. For some examples of how the query would work, so here you have a curl command with the query of our API. You can filter by country or you can also put your own polygon into it and by days and you get a JSON format with the values. You can then plot this. This is for example just February in Germany, February 2019 and the average for each day. So you can easily make comparisons for example between measurement days. Then you can plug it into your favorite visualization here. We did an example with backgl. That just kind of looks cool because you have the 3D effect and you see the different areas in Germany and their emissions at that point when the flyover was taking place. Well, we still have a lot of challenges to resolve. We realize during the process obviously that this is way too much data. We somehow need to reduce it. We are currently playing around with geospatial indexing systems but maybe somebody else has a better idea how to do it. We also need a long-term host. Currently we got some cred from Amazon from AWS and we also have some universities that are interested in hosting our project but we still need a solution that has enough power for long term as well. And of course we need to import more emission data more of the other products from the satellite. So this is our story of getting involved in public interest take. For us from that point of view the real interesting takeaway was that there is a lot of interesting data out there but usually it's just very difficult to access and we probably need a lot more people to get working on this infrastructure to build actually infrastructure that comes before building a tool or visualization. And so the easiest way if you want to get started in this is that you just get started with our project and have us developing or use our product. First off you would go to emissionsapi.org check the examples, check what we are doing. You can find us on GitHub obviously and hit us up on Twitter to just know what we are missing and what you would like to have in this product. I want to thank also our sponsors prototype fund that is part of the Open Knowledge Foundation in Germany and both of this is sponsored in the Empire the German Ministry of Education and Research. Thank you very much.