So good afternoon to everyone. I'm glad that I can be here. And it's really good to continue after Mr. Kakmaji because my presentation is, let's say,

continuing with the topic, which was introduced by the previous speaker. And I would like to tell you about the project that was mentioned in the presentation, in the previous presentation. And it is because it was done between our company, CGI,

where I work, and between Technical University of Ostrava. So maybe for the beginning, I'm quite used to challenge the situation that people doesn't know my company. So firstly, I just shortly tell you

that CGI is quite big IT company operating worldwide. And it has many offices in Europe, especially in Western European countries. And here in Czech Republic also has three offices in Prague and Ostrava, and also one in Bratislava and in Budapest.

As you can see, approximately 800, I'd say, IT gurus and guys are working in Prague, mostly, and in Brno. And we are doing projects for not just public sector, but also for another private companies,

and not typically for end users. So that's the reason why the people usually don't know about our company because we are not usually providing services to end users. In my team, even if we are big corporate, even if we have many offices in Europe, so here in Prague,

we have approximately around 20 people. Most of them work in Olajevic district, where is the location of USPA, European Agency for the Space Program. And they are focusing on public regulated services, so services which are typically used by special entities, which

has access to coding services provided by European Navigation System. And in my part of the team, which is like five people on the other side of the screen, we are doing our observation projects, which are using Copernicus data sets, and they are usually supported by ISA funds.

And as the Mr. Kaczmarek told you a few minutes ago, we had a project which ended at the end of the last year. Quite complicated name, Earth Observation Automated Monitoring Open Platform. But the solution or the aim of the project

was to develop a cloud-based platform to process satellite images, mostly automatically, and to evaluate them and detect, let's say, changes which happened during the time of interest. And we are cooperating with the Technical University

of Ostrava, and their responsibility was to develop for use cases, which were, let's say, quite in detail introduced by Mr. Kaczmarek. On our side during that project, which took two years, we decided to implement two of them, cloud detection

and displacement detection. And the reason was that we evaluated these use cases as the most valuable for potential users and customers. And what is good to say that because we were cooperating with the university, the results are, let's say, full of details about the data which were used, so Sentinel-1,

and they are full of many statistical metrics, which are typically for a long temporal series analysis. So these data are sometimes really hard to interpret for users which are not involved in geospatial data

processing, or they are, let's say, not fully, or they are not typically used to, let's say, evaluate Earth observation analysis. So for that reason, I will skip this slide because it's introducing the technique for the displacement

detection, but it was mentioned a few minutes ago. So for that reason, we decided to, because it was not possible to do it in the original project, for extra money from different fund, and to build a special web

application which is using the results from displacement detection for, let's say, insurance sector. And why insurance? Because USPA presented in their market report at the end of the last year that insurance and finance are the sectors which will grow as the fastest

in the following 10 years, and they will use much, much more, let's say, spatial products and Earth observation products in following years. So during that project, we want to develop a web front-end application which

is easy to use, even for people which are, let's say, completely out of geospatial work. And if it's possible to deliver them quite complicated products from the displacement detection analysis in the simplest form, in the form which is easy to understand today,

and in the form which everyone can understand quite easily. So we built that solution on mostly, or especially just on pre-open source technologies, as well as SQL, Postgres, GeoServer, and so on. So I will show you.

I hope I will have a few minutes at the end of the presentation, the application. And that's how it looks. It's simple web application with map window, which is in your full screen mode. And you can easily navigate through different parts of the city, for example.

And you will see in colors different evaluation of stability, which is based on persistent scatterers, which were introduced by Mr. Kaczmańik in the previous presentation. For those who are interested in technology, so the architecture design is composed

from two parts, front and the back end. And it's a fully cloud-based solution. So we are using Microsoft Azure as a cloud service. On top of that, we are using Container Orchestration System, Kubernetes, with Docker containers, which

are isolating applications, which are running fully independently. So they can be replaced, updated, let's say, smoothly. And we can be sure that we will not break the application or the whole solution if we do some small changes.

So this is quite typical microservices solution. And the front end application is based on JavaScript and two libraries, React and Leaflet, to visualize the results in the map window and in the browser.

And we are using Peplog as identity and access management solution, which is ready to use. So it's just necessary to set up everything correctly. And you can secure your solution. And for example, protocols, different customers, let's say, different login settings, roles,

and access to your data. So good to mention. And I would like to say that we are not just build the solution on Sentinel-1 data on the displacement detection analysis,

but we are also using another third-party services, which are free to use. Here in Czech Republic, I'm speaking about, let's say, services which are provided by Czech office for surveying, mapping, and catastrophe. So I'm speaking about autocode maps, catastrophe maps,

and also some quite high-value service. It's search geocoding service, which you can use if you want to look for special addresses and their location in the config. So combination all of these services, we build the solution. And of course, Sentinel-1 data have special resolution,

around 5 by 20 meters. So for, let's say, each building in ideal work, you can get one reflectance for radar waves. But also, not all the time, the situation is perfect.

So for some buildings, you will get more, let's say, reflections. But for some, you will not get almost. You will get one or even none. So it depends on the situation. So you cannot say in advance how many, let's say, points and time series observations

you will have for some specific location. But it's good to say that if you are speaking about the insurance sector, what is typical for them and what is quite good for people to interpret the data is to combine them with something what they know really good.

And that's typical colors to maps, because they are, let's say, related to ownership rights and also to insurance companies which are providing to users and owners, let's say, services to their properties.

So we combine all of this information and create, let's say, open post-processing algorithm for time series in the form of points. We merge all of the data together. We created open evaluation metrics,

reclassify the data, and combine them with the cluster maps to make them as easy to understand as it is possible. All of this solution is, again, was implemented into our platform in the form of a special Docker

container, which is based on QGIS and Python, mostly PyQGIS and GeoPython. And we are using also GDAL to convert this data to a form and store it in Postgres to your database with Postgres extension. And after that, typical workflow

published there via GeoServer as services. So the typical operation scenario for user is just to log in, search for the desired locality, for example, manually or by address,

because we are using the geocoding service. Check the stability, check some additional information if they are provided, and export the results or they can be happy just to know what is happening there or not. So I will show you a short demonstration. And I have to log in into the application.

So give me a second. This is just a demo. So we don't have so much data right now in the application for the demo mode. But for short demo, it's enough.

So at the beginning, after login, for each user, there are, let's say, specific locations which are already evaluated and already calculated, because maybe it was not mentioned,

but the processing of long temporal series of Sentinel-1 data sets, it takes usually days or weeks, because we are speaking about terabytes of data sets. So if you will have the data, you will have to prepare them in advance to enable the users to reach them just as results.

So when I will go to some location, so right here, I have the Valeshovich district. And I can go, let's say, just by simple navigation to some specific part. And here is quite new, shopping mall. So after clicking on the map, because it's,

let's say, a little bit pink color here, so it should be something what is not so stable. So I will see that the stability value is around 80%. That's based on our classification. But most of the points which are composing the area or the reflectance information which

we have from displacement detection analysis are stable. So I can check it not just by the color of the map, but of course, I can also switch to different modes. So I can go directly to persistent scatter points. I can also, for example, switch off the cadastral map.

And I will have, let's say, the point layer which is representing the reflectance. I will see, I can see that I have many green points which representing that nothing is happening there, but also some which are yellow or, let's say, kind which representing direction of movement up or down.

But because we are speaking about points, it's not possible to say they are moving up and down, let's say horizontally or vertically precisely because I have measurement just in the line of sight of the radar and the reflector on the ground.

So if I will check some points which are yellow, I can see that they are really like changing the distance in the time series. And they are farther and farther from the satellite during each observation which can be expressed that the point is like going down typically.

And if you will combine all the information which are here for the one building in one footprint of the building, you can say also, if it's really like going down, the movement is horizontal and it's typically to see it in the polygon layer. Maybe I haven't mentioned that, but if I go here,

I will see, you can see that the displacement direction is down. So by this, let's say simple check, even like in the simple form or in like deep mode, if you can evaluate each point independently,

you can evaluate the whole area and say, it's something happening there. If there is some potential risk for some insurance company or the owner that the building can be, let's say affected by some accident, or it should be done something to save the money

of the owner of the building. For example, for another buildings, you can see that nothing is happening there. And if some movement was detected, it's usually under one millimeter per year. That's let's say the ratio of the precision of the method. So thank you for your attention.

This is a short demo. If you have any questions, you are free to ask. Thank you. For the last presentation. So are there any questions there in the audience?

There's buildings in the building. It's a good question. Unfortunately, the sample dataset doesn't contain this area, but we are right now working on the evaluation of the whole Prague, four, five years, like from 2017 to 2021.

And yeah, we will have the results on say in one or two weeks. But typically what we have evaluated, most of the Prague is stable. Just some small parts are unstable and they are usually located close to the river.

So, but I cannot say, or I cannot show you the results for that building where we are right now. Okay, so thank you.