Visualizing climate risks for disaster reduction and climate resilience programs – Interactive open-source tools for analysts and decision makers to utilize Earth observation data
This is a modal window.
The media could not be loaded, either because the server or network failed or because the format is not supported.
Formal Metadata
| Title |
| |
| Title of Series | ||
| Number of Parts | 351 | |
| Author | ||
| Contributors | ||
| License | CC Attribution 3.0 Unported: You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal purpose as long as the work is attributed to the author in the manner specified by the author or licensor. | |
| Identifiers | 10.5446/69179 (DOI) | |
| Publisher | ||
| Release Date | ||
| Language | ||
| Production Year | 2022 |
Content Metadata
| Subject Area | ||
| Genre | ||
| Abstract |
| |
| Keywords |
FOSS4G Firenze 202265 / 351
1
7
13
22
25
31
33
36
39
41
43
44
46
52
53
55
58
59
60
76
80
93
98
104
108
127
128
133
135
141
142
143
150
151
168
173
176
178
190
196
200
201
202
204
211
219
225
226
236
242
251
258
263
270
284
285
292
00:00
Reduction of orderDecision theoryOpen sourceInteractive televisionComputer programProjective planeInformation engineeringPresentation of a groupBitContext awarenessExecution unitData managementState observerComputer programmingGoodness of fitDivision (mathematics)WebsiteComputer animation
00:34
InformationSatelliteBitContext awarenessSoftwareBuffer solutionOffice suiteProjective planeDenial-of-service attackComputer animation
01:12
Slide ruleComputer programVulnerability (computing)InformationReduction of orderDependent and independent variablesSystem programmingMathematical analysisPoint cloudLambda calculusFunction (mathematics)Process (computing)Open setCubeScalabilityPrice indexService (economics)Data analysisComputing platformDatabase normalizationSatelliteComputing platformInformationCubeOpen setObservational studyRaster graphicsLatent heatServer (computing)Term (mathematics)Execution unitSelf-organizationDifferent (Kate Ryan album)Order (biology)Multiplication signOffice suiteElectric generatorMathematical analysisPhysical systemModal logicPersonal digital assistantVulnerability (computing)Revision controlConnected spaceOpen sourceSingle-precision floating-point formatContext awarenessWeb serviceSubject indexingPrice indexLaptopWeb 2.0Time seriesDuality (mathematics)Potenz <Mathematik>Computer animation
04:52
Slide ruleService (economics)Image resolutionOperations researchMathematicsLocal ringMathematical analysisServer (computing)Point cloudWeb browserSoftwareInterface (computing)Directed setCollaborationismWorkstation <Musikinstrument>CubeOpen setSoftware bugComputing platformProduct (business)Library catalogComputer fileVisualization (computer graphics)Series (mathematics)PixelSatelliteCase moddingProcess (computing)Basis <Mathematik>BefehlsprozessorSingle-precision floating-point formatOpen sourcePressure volume diagramMathematical analysisInterface (computing)Software bugLaptopSubject indexingInternet service providerServer (computing)Service (economics)Prisoner's dilemmaDatabaseHazard (2005 film)Open setInformationOpen sourceProduct (business)CubeMultiplicationTerm (mathematics)Process (computing)Case moddingPixelFunction (mathematics)TesselationAsynchronous Transfer ModeRow (database)Computing platformMultiplication signDifferent (Kate Ryan album)Electric generatorImage resolutionTime seriesDebuggerCartesian coordinate systemConnected spaceRaster graphicsProjective planeRaw image formatSurfaceComputer animation
07:38
Software frameworkVulnerability (computing)Hazard (2005 film)Computing platformComputing platformInformationDisintegrationSystem programmingPoint cloudSatelliteCubeOpen setOpen sourceMathematical analysisElectronic visual displayAreaLocal ringDivision (mathematics)Arithmetic progressionMaizeService (economics)Server (computing)Open sourceServer (computing)Form (programming)Office suiteBoundary value problemIncidence algebraShared memoryComputing platformInformationType theoryGeometryPrice indexPrisoner's dilemmaFunction (mathematics)PolygonResultantVulnerability (computing)Hazard (2005 film)Mathematical analysisInstance (computer science)SubsetMedical imagingFront and back endsDivision (mathematics)Subject indexingService (economics)BitCASE <Informatik>Raw image formatLevel (video gaming)AreaDebuggerCubeOpen setSystem programmingVisualization (computer graphics)SharewareSurfaceSystem administratorAutomationCodeComputer animation
11:22
Mathematical analysisOpen sourceMathematical analysisTesselationLevel (video gaming)InformationSystem administratorOpen sourceMereologyNetwork topologyAreaSystem programmingOperator (mathematics)MultiplicationStatisticsServer (computing)MultilaterationConnected space
12:57
Internationalization and localizationInstance (computer science)Computer fileLocal ringConfiguration spaceInformationIndependence (probability theory)DebuggerFluid staticsLevel (video gaming)Context awarenessServer (computing)Key (cryptography)Fiber bundleTesselationDifferent (Kate Ryan album)Medical imagingCuboidElectronic mailing listPresentation of a groupFront and back endsLink (knot theory)Zoom lensAdaptive behaviorReverse engineeringCodeComputer animation
14:10
SharewareMaizeEmoticonSharewareMultiplication signComputer animation
14:30
Compilation albumGEDCOMComa BerenicesOptical disc driveSupersonic speedComputer fileSharewareCategory of beingLevel (video gaming)Prisoner's dilemmaInformationCubeIntegrated development environmentOpen setRepository (publishing)Configuration spaceStructural loadBoundary value problemMedical imagingMultiplication signFront and back endsVariable (mathematics)Electronic visual displaySource code
16:11
PlastikkarteRule of inferenceCorrelation and dependenceDrop (liquid)Schmelze <Betrieb>SineAutomorphismSolitary confinementSalem, IllinoisInterior (topology)Computer fontTorusBit rateGame theoryInfinityMassVertex (graph theory)Row (database)GEDCOMForceAreaMathematical analysisCone penetration testTask (computing)Multi-agent systemSuite (music)Conservation of energyThresholding (image processing)InformationAreaSharewareFront and back endsDebuggerCubeStatisticsElectronic mailing listOpen setPrisoner's dilemmaResultantSystem administratorMathematical analysisDifferent (Kate Ryan album)Boundary value problemLink (knot theory)Point (geometry)Line (geometry)Buffer solutionCASE <Informatik>PolygonCategory of beingType theoryReading (process)Multiplication signMathematicsImage resolutionLatent heatTotal S.A.Raster graphicsString (computer science)Vector spaceElectronic visual displayTerm (mathematics)Computer animation
Transcript: English(auto-generated)
00:00
Thank you. Good morning, everyone. As the presenter said, my name is Jorge Martinez. I'm working on the emergency division, the GIS unit of the World Food Program here in Rome. So the trip was easy, two hours, quick. And I'm collaborating with the RAM, Research Assessment and Monitoring Division, which is the Climate and
00:20
Observation Unit. So this is like a side project between two divisions. And the global program manager is Amit, and Anthony is in charge of the Earth's observation and data engineering. So let's start to give a little bit of context of why we have developed this software, or the
00:42
project itself. And what happens is that constantly, we're receiving requests about emergencies like floods, tropical cyclones, earthquake information. And usually, those requests come from country offices and governments, which has low income and no strict or
01:00
well-defined access to the information regarding satellite information, regarding tropical cyclone wind buffers, how many people were affected, and so on. So we come with multiple challenges. And one of the challenges is that Earth observation data requires technical skills to access and use it. For example, the data that comes from GDAC requires an
01:22
API, which the way we do requests for the data is different from the data that you need for earthquake information. Or the rainfall information, or the rasters come from satellites, requires a different way to access it. The second one is governments in low and middle countries have limited resources.
01:41
As I mentioned, not all the countries have enough budgeting for having a GIS unit inside the agency, or even inside the country offices. Or even the resources in terms of infrastructure, like a server, to access all the information automatically. It's not easy for everyone.
02:02
And data risk and vulnerability are spread across systems, which means there's not a single source of information to access what you need or what you want. For us, it remains like different opportunities, which means technical assistance to governments is a key modality to WFP Works.
02:22
WFP as an organization aims to connect with different governments in order to improve the logistic and improve the performance of what WFP usually does. Climate risk analysis and vulnerability info improves targeting, so you can, with a better analysis and with better tools designed, you can specifically go to the
02:42
specific place to reach the specific people, amount of people. And faster access to information, which is like it transformed to less impact, so you can know specifically what you want. Based on that, the challenges and opportunities, we come with three main milestones.
03:01
So the first one is to collect, visualize, and generate time series analysis for different climate indicators. So we can pick drought, vegetation, and rainfall. There's a need in terms of infrastructure, like a single platform that would work for multiple countries, because many of the data is like a country-dependent and
03:22
context-dependent. For example, usually in the Caribbean, the data that is mostly requested comes from tropical cyclones and rainfall, while in some other countries, drought is the main indicator for dual analysis or the data that they need. And connection to external data sources, which means that not all the data that WFP generates is going to be
03:43
conserved and visualized inside WFP. It is generated inside WFP, which means that we have other data sources outside, like USGS, GDACs, armed conflict information. We can connect that information and visualize and do the analysis. So for the first one, we come from a platform that many
04:05
people, many have heard of it. It's Open Data Cube. So Open Data Cube is like a tool that is designed to collect index information in order to run analysis inside the same platform. It's like an online web tool. So people get the data.
04:21
You can easily get the information you need regarding satellite information. And then it's indexed data. And then you can use it for running analysis, or you can use it to serve that information as a web service. It has its own APIs. This is like the main goals, why we decided to use Open
04:41
Data Cube, which is provide an internal platform for analysis. It has an inside Jupyter notebook array. And the data is accessed publicly. In terms of the infrastructure, we collect data from NASA, from Climate Hazard Center. We collect data from Surveyor MODIS.
05:03
And then we get that information inside Open Data Cube. Let me use this. OK, we get the information. And we index everything inside the Open Data Cube. And this data is going to be served for different business
05:21
and application services. Inside WFP's database and aims, the government one and the one that I want to talk about after this is Prism. Also, we can connect to S3 services. We can get the data. Open Data Cube has a plug-in called Data Cube OWS.
05:40
So you can serve that information as an OGC API, which supports WMS and WCS request for rasters. And we can provide analysis interface, which people inside, there's a Jupyter notebook, so people can access and do the analysis if they need that. In terms of infrastructure and the data that we manage,
06:03
we're using rainfall information, vegetation information, which is NDVI, temperature, which is land surface temperature, and the anomalies, which is analysis made for that. We currently have 1.2 terabytes of raw data and 1.5 terabytes of derived data.
06:20
This data is accessible through OGC requests. So if you have a platform like a RJS online that accepts WMS request, you can connect and visualize the tiles. And this is an example of the derived data that we use, which is, this is called a filter,
06:40
and basically the data that comes from MODIS is not good enough for me, like analysis, so we use a filter and process, we process the data and generate the output, which is a filter information. This, for MODIS, is like the biggest challenge for us, because the tile has like a 1200 times 1200 times
07:01
like 98 pixels, which is like a beak and takes a while. And we filter three projects with one kilometer resolutions, and we use IWS Fargate for this. So yeah, this is like the data that we use, which is, we fulfill like this first milestone, which is regarding to collect, visualize,
07:21
and generate time series of data. We take data from different data sources, most of them are rasters, and then we analyze as we process it, and the data is gonna be served. But for the second milestone, which is the platform for multiple countries and connection to external data sources, we have developed a front-end tool, which is called PRISM.
07:41
So the idea of the PRISM is that you can collect data not only for regarding the hazard from the open data cube or the humanitarian data cube, which is mentioned that we got, but we can get information from the government, which is regarding vulnerability or exposure data. Not necessarily can come from the government, but also can come from like directly in raw data,
08:02
but you can use like forms, like a code forms or open or ODK forms and put it into PRISM so you can run analysis and then you can measure what you, yeah, fulfill the goal that you want or what you need. It's designed as a tool for government institutions. It's open source, so each government can,
08:23
you can set up your own instance of PRISM so you can define what is the country, which are the layers that you need, and what's the information you want to displace and which type of analysis that you want to generate. So, and yeah, it can catalyze institutional data sharing because we can, the analysis that you can do
08:41
inside open data cube or that you can do on PRISM, you can download and you can share it freely. So yeah, this is like the first data source that we use, which is like the output from the open data cube. It's like the results and the analysis are fed inside the platform. It's like a visualize as a OGC service WMS request
09:03
and we got information that is placed previously like the rainfall anomalies, vegetation index, surface temperature, so you can visualize it on the platform. We can also have the results from forms collected by the government or like raw data information is classified by admin level.
09:21
In this case, it depends on the country. Some countries use admin level three, sometimes use admin level two, and then we just get that information, get the results or indicators, and then we just put into the platform so they can be visualized. And finally, emergencies, in the emergency division,
09:40
we have a platform called Adam, which is an automated disaster alert system. Basically, it's when there's an earthquake or when there's a tropical cyclone, it alerts the people subscribe that that incident happened. It happened in a given area, how many people were affected, how close are the country offices
10:01
or the warehouses in that city. Yeah, we have integrated that. We got the information regarding the boundaries and the polygons, and then we just serve that through GeoServer as the WMS request and then Prism just, it takes charge of visualization. The WMS request, the GeoServer takes advantage
10:23
that we can get also the raw data itself, like the WFS, and then we can just run analysis for example, how many people were affected. I can show that on the demo in a bit. And this is open source solution. It's fully open source. We have the front end,
10:41
which is built on TypeScript, React, and MapLiver. The API, we have a backend API, which it takes charge of downloading the data and running the analysis that are needed, and their outputs are returned to the front end. It's built on Python and Flask. You can use Docker as a Docker container for the Docker image for that,
11:01
and we use GDAL for running the analysis. And the data services provided, if you can connect most of the data services that accept OGC requests, we use PostgreSQL for managing the data and GeoServer for serving it. Currently, there are like 20 contributors
11:20
and it's MIT license. Yeah, it's a flexible tool. Basically, you can connect multiple data sources. As I mentioned, if you have anything that serves, that treats, I mean, that accepts WMS requests, you can get the data out of the tiles. Easily overlap and explore data layers. You can have WMS data and you can overlap
11:42
with information from the government so you can make a better analysis. The Geospatial API, you can get, it's a sonarly-statistic aggregation. You can see for each admin level what is the rainfall information or the trial information. You can intercept layers, which is gonna be,
12:02
I'm gonna demo it regarding the tropical cyclones and handles compute-heavy operations. This is the part that we are currently improving, which is the alerting system. Users want to know when something, if a given admin tree gets a lot of,
12:20
it's gonna have a lot of rainfall. It's gonna rain a lot in a given admin tree so you can alert the people subscribed to notify that in that area specifically it's going to rain a lot. Detect new layers and launch analysis. Yeah, exactly. When the alert happens, you can run the analysis that you need to measure how many people were affected,
12:41
which are the possible areas affected and so on. Yeah, notify and send the users. Mostly data sources are like a GeoServer, ODC, GeoJasons. It's usually classified by admin level. The key benefits of Prism is that the front-end is serverless,
13:01
you can just, it's built in React so you can make the bundle and then you can serve as a static file, or yeah, as a static HTML and JavaScript using NGINX in the server, you can deploy to search or reversal as an HTML. Sure, back-end, independent front-end, you can have different instance of Prism by country.
13:23
So you can have two files, I'm gonna show it right now. You can have two files, one is a layers which specify the list of layers, and the second one is the information of the country, the bounding box of the country, the zoom level that you want the data, which are the tiles that you want. Configuration, no coding skills required,
13:40
as I mentioned, two files, layers.json, layers and the Prism configuration file. Adaptable to local context, we're working on the localization, so you can, it's gonna be reachable for like a country, at a country level. Yeah, no permissions required in fully transparent. Everything is on GitHub,
14:00
so you can get down to the file, you can build your own instance, you can install React, you can run the Docker images and so on. So yeah, these are the links. I'm gonna close this and show the, what, the demos.
14:23
Okay, thank you. Can I just, yeah, which time is less? Up, okay, perfect. So this is like the repository, this is the Prism front-end repository.
14:41
As I mentioned, it's built on React and this is like the API, the API has a Docker image and it's built on Python and Flask. The main thing that I want to show here is like the configuration. So these are the multiple countries that we have Prism deployments, not necessarily every country here
15:02
is like official deployment of Prism, it's just like for demoing and displaying. And as you see, let's pick one, okay, let's pick Tajikistan. There's like two main files. The first one is layers.json, which contains like the configuration of the boundaries that are gonna be used.
15:23
Also, if I search for WMS, this is like the data that is reading from Open Data Cube. And the other one is the Prism configuration, which contains like the all requests, the map, how we're gonna center the map, which is the center of the country and the category. So you can configure the rainfall information
15:40
is gonna be on a specific category. So what happens is that internally, when you run the, when you deploy Prism, it creates like a dictionary per country and then based on environment variable of that country, you can, Prism will try to render all the layers
16:00
and all the configuration according to that environment variable. So it's basically like a dictionary, Prism loads, takes the country you need, loads the layers, loads the configuration, loads everything that it needs, it's needed. So you can see here, there are like two main deployments. First, this is like a demo of Mozambique. And this is like a demo of, oh, can it change?
16:23
Okay, so yeah, I can, there's gonna be like a demo of Colombia, but I cannot see the link, it's like a wrong link. But in general, it's a different country, it's a different list of categories and it's a different list of layers that's gonna be displayed. So here I'm displaying the NDVI information that comes from Open Data Cube,
16:41
but I can display here information regarding rainfall. Yeah, rainfall information. But I can also read vector data, which is that I have collected here.
17:01
Okay, sorry, resolution. The tropical storms data. So you can here visualize all the tropical storm data collected for Mozambique. Let me show here.
17:21
And yeah. So this information regarding tropical cyclones, there are like two wind buffers, there's like a list of points where the episodes have been generated. And this is the light string, which is the path of the tropical storm. And Prism has the ability to run some analysis. So if I want to run this type of analysis,
17:40
which is exposure analysis, takes a while, sorry.
18:05
I cannot change, change the link for this. Just remove this. So yeah, this is the result of the analysis. What it does is that the front end of Prism connects to the backend, gets the information regarding the tropical cyclone,
18:22
all these GeoJSON, which is the wind buffers, the points and the lines. We take the boundaries that are also fed from the front end. So this is like the boundaries. And then we just initially do the intersection between the admin boundaries, in this case is admin tree, which each of the polygon wind buffers.
18:41
We take the intersection, and then we pass that to the sonar statistic analysis, which is the raster of population. Then we can see for each polygon, we can see how many people were affected. In this case, in this example, you can see the same city, but with different wind buffers intersection, the people that were affected.
19:03
It has another type of analysis, which you can measure the rainfall information, and then you can compare with other types of data, for example, total population, and then you run the analysis. So you can measure, okay, in this area, there's gonna be probably a precipitation of high precipitation, and the population affected is gonna be this.
19:21
So you can do some sort of like a preparedness and then you can allocate resources to that area, a specific area. So that's more like in terms of like efficiently. So yeah, it's not like a different, it's not like all the country or the whole city, it's just a specific area. So you can run the analysis and allocate the resources to that area.
19:42
Yeah, I'm thinking I'm out of time. So I'm gonna stop for this demo. If people are interested, I can show the demo for the other countries, for Mozambique and, oh, this is Mozambique, but I wanted to show them of Colombia and Cuba. Yeah, thank you.