GlobWetland Africa Toolbox: Implementing Earth Observation based Wetland Monitoring Capacity in Africa
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 | 295 | |
| Author | ||
| Contributors | ||
| License | CC Attribution 3.0 Germany: 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 | 10.5446/43297 (DOI) | |
| Publisher | ||
| Release Date | ||
| Language |
Content Metadata
| Subject Area | ||
| Genre | ||
| Abstract |
| |
| Keywords |
00:00
Projective planeSatelliteSimilarity (geometry)BuildingChannel capacityLecture/Conference
00:29
Channel capacityIntegrated development environmentGrass (card game)Focus (optics)BuildingProjective planeChannel capacity
00:46
Service (economics)Total S.A.Type theoryTerm (mathematics)Traffic reportingData storage deviceSatelliteService (economics)Water vaporArmChannel capacityWebsiteMereologyClosed setLimit (category theory)Computer animation
02:04
Prime idealSatelliteSoftwareOpen sourceChannel capacityOpen setFreewareObservational studyComponent-based software engineeringChannel capacitySatelliteMultiplication signOpen sourceComputer programDiagram
02:59
SatelliteOpen sourceSoftwareChannel capacityFreewareWater vaporParameter (computer programming)Mathematical analysisTerm (mathematics)Inclusion mapPersonal identification numberDrum memoryPressureCollaborationismSpacetimeProjective planeCartesian coordinate systemProcess (computing)Presentation of a groupBitLibrary (computing)Point (geometry)PreprocessorProduct (business)SatelliteComponent-based software engineeringQueue (abstract data type)ChainFile formatGraphical user interfaceChannel capacityBuildingMathematical analysisOpen sourceDiagramComputer animation
04:03
User interfaceOpticsLibrary (computing)Data managementProcess (computing)Grass (card game)Computing platformProcess (computing)Software frameworkFreewareOpen sourceEndliche ModelltheorieScripting languageAlgorithmScientific modellingInternet service providerMusical ensembleProcess (computing)Software bugOpen sourceUser interfaceAlgorithmSatelliteWorkloadCodePreprocessorProjective planeCartesian coordinate systemParameter (computer programming)Modulare ProgrammierungChainDescriptive statisticsWave packetPlug-in (computing)Greatest elementMultiplication signMathematical analysisQuantumFunctional (mathematics)Component-based software engineeringInternet service providerSimilarity (geometry)Graphical user interfaceGrass (card game)Scripting languageSoftwareComputing platformRevision controlLibrary (computing)outputSoftware developerEndliche ModelltheorieIntegrated development environmentFreewareExtension (kinesiology)UsabilityPoint (geometry)Sound effectOperator (mathematics)Information systemsInformation securityCircleDifferent (Kate Ryan album)Function (mathematics)Electronic data processingDataflowSoftware frameworkMereologyTask (computing)Installation artComputer animation
09:39
AlgorithmInternet service providerSoftwareImage processingSatelliteGraph (mathematics)Software frameworkUser interfaceProcess (computing)Bimodal distributionProcess (computing)Lemma (mathematics)outputMedianLink (knot theory)Twin primeMusical ensembleResonanceRankingHill differential equationData typeMenu (computing)Type theoryFreewareInclusion mapAdvanced Boolean Expression LanguageCommon Intermediate LanguageComputer wormOpen sourceInflection pointMotif (narrative)SpacetimeGraphical user interfaceData modelPhysical systemAlgorithmTask (computing)Installation artProcess (computing)Different (Kate Ryan album)User interfaceEndliche ModelltheorieSoftwareSoftware developerPlug-in (computing)Electronic mailing listComponent-based software engineeringGrass (card game)Computer iconDisk read-and-write headUniqueness quantificationCategory of beingDirectory serviceInformationConfiguration spaceLibrary (computing)Computer fileGraph (mathematics)Internet service providerParameter (computer programming)Repository (publishing)1 (number)Semiconductor memoryGraph (mathematics)Heegaard splittingoutputGraphical user interfaceSoftware frameworkSet (mathematics)TesselationField (computer science)BitFunction (mathematics)Projective planeMereologyCartesian coordinate systemWave packetMaterialization (paranormal)Scripting languageStandard deviationExpert systemFocus (optics)Type theoryUniverse (mathematics)Product (business)DataflowGauge theoryForcing (mathematics)NumberOcean currentProfil (magazine)Speech synthesisSelf-organizationLine (geometry)DemosceneQueue (abstract data type)E-bookComputer animation
15:16
Execution unitProduct (business)Proof theoryType theoryTexture mappingFitness functionWater vaporTemporal logicPressure volume diagramSatelliteSystem identificationComputer fontSystems engineeringSeries (mathematics)MultiplicationPopulation densityWebsiteEstimationComputer-generated imageryVariety (linguistics)Exploit (computer security)OpticsType theoryNumberProduct (business)WebsiteTime seriesSatelliteImage resolutionState observerDifferent (Kate Ryan album)AreaMappingMultiplication signNeuroinformatikClassical physicsWorkstation <Musikinstrument>Computer animation
16:03
Inclusion mapSurfaceSurfaceMultiplication signCovering spaceSocial classAreaEvoluteArtificial neural networkComputer animation
16:27
Temporal logicPressure volume diagramBounded variationSurfaceSatelliteFrequencyMaxima and minimaSeries (mathematics)Water vaporTerm (mathematics)Absolute valueGraph (mathematics)Graph (mathematics)EmulatorProcess modelingInformationSound effectDependent and independent variablesMathematicsImage resolutionStructural loadMetreOpen sourceFrequencyExtension (kinesiology)ConcentricPrice indexPoint (geometry)AreaMeasurementLatent heatCurvatureMultiplication signMaxima and minimaSurfaceOpen setInformationEndliche ModelltheorieWater vaporComputer animation
17:22
Data structureRegular graphInformationForestTime evolutionImage resolutionPopulation densityDirac delta functionPressure volume diagramTemporal logicTexture mappingComputer fontSpeciesComputing platformOpen setFreewareOpen sourceProcess (computing)AgreeablenessHeat transferPoint cloudOpen sourceUser interfaceSoftwareUsabilityTime seriesScripting languageRevision controlInternetworkingConnected spaceProcess (computing)NeuroinformatikServer (computing)ResultantoutputMultiplication signVirtual machineFreewareSoftware developerPoint cloudComputer animation
18:54
Internet service providerComputing platformPlug-in (computing)Projective planeCondition numberDefault (computer science)Metropolitan area networkSoftwareInstallation artWebsiteRevision controlForm (programming)Discrete element methodProcess (computing)Function (mathematics)Flow separationAddress spaceOpen sourceMathematicsRepository (publishing)Lecture/Conference
Transcript: English(auto-generated)
00:07
Good afternoon, my name is Raspav Guzyński, I work for a Danish company called DHI-GRAS and we work with satellite remote sensing and just to avoid any confusion we are not
00:20
connected to the GRAS GIS project, we only use it despite the similarity in the name. So we're talking about one of the outcomes of a project which we have, the project is called Globe Wetland Africa and the focus is on increasing the capacity
00:40
so capacity building for wetland monitoring in Africa using remote sensing data. So firstly why are we focusing on wetlands? Well they are one of the very important ecosystem types, here it is expressed in monetary terms where the wetlands provide over 50 trillion
01:02
or close to 50 trillion US dollars worth of ecosystem services, mostly through water filtration or flood protection, but they're also very important in biodiversity terms or general ecosystem functioning terms. So because of this importance there's this international
01:21
convention on wetland protection, it's called RAMSA, and the most important global wetlands are within this RAMSA convention, they're called RAMSA sites, and close to half of those RAMSA sites are based in Africa, so close to half of worldwide or globally important wetlands are based in Africa. And as part of this RAMSA convention the countries in which
01:44
wetlands are located have to do some reporting, for example wetland inventory, but due to limited capacity and resources many of the countries have not yet achieved their
02:01
inventories or have not delivered their inventories to RAMSA. So there is a need to increase the capacity of the stakeholders to fulfill their obligations towards the RAMSA convention, and with the new satellite sensors launched over the last five years, mainly through the Copernicus program, so the Sentinel
02:25
satellites, but as well other sensors, Landsat for example, now there is a lot of free and open earth observation data that is very suitable for this purpose of monitoring wetlands. At the same time there are many open source tools, as we all know, because we are here for
02:43
this reason, for GIS the tools are as well quite mature and suitable for the purpose, so kind of what is missing is putting those two components together to help the countries fulfill their obligations. So this is where the, well this is where this global in Africa
03:03
project comes in, it's sponsored by the or funded by the European Space Agency, and it was the work is done in collaboration with RAMSA. So the tool which is used for this capacity building is called global wetland Africa toolbox, it's an open source toolbox
03:25
which was developed during the project and it focuses on six main wetland applications which are listed here, and later on towards the end of the presentation I'll go through them in a bit more detail. But the point is that for each of those wetland applications there is a
03:42
processing chain within the toolbox going from the very beginning, so from opening the satellite data in the native format that is downloaded, for all the pre-processing steps, analysis steps, to deriving the final product needed for those applications.
04:04
The toolbox uses QGIS as it's a major component, so the QGIS serves as the graphical user interface of the toolbox, and then within the QGIS we have this library of workflows, so for each of
04:25
the wetland applications there is one or more workflows which you can see at the bottom of there, they are basically step-by-step instructions guiding the users through the whole processing chain. And for each step in the workflow there is the panel on the left
04:46
where there is the description of why this step is performed, why it's important, and there is guidance about which parameters to choose, which inputs and outputs should be set.
05:05
So QGIS is, you could say, the main platform of the toolbox, but the actual processing, so the analysis of the data, the pre-processing of the data, is based on algorithms coming from many other open source projects. You're probably familiar with most of them, but maybe some
05:23
people don't know SNAP, which is the Sentinel application platform, which is a toolbox developed specifically for analyzing the Sentinel satellite data. BEAM at the bottom is an older version of SNAP. We have to use it because we are also working with historical data from
05:43
NVSAT satellite and not all the functionality was transported to SNAP. As well, there are custom R scripts and Python scripts which were written to fill in the gaps where existing algorithms couldn't perform the function that was required. So we have all the
06:07
kind of algorithms available from the software, but everything is available through the QGIS graphical user interface and through the workflow. So it's very standardized and the user doesn't have to go to the different software to get the different functionality.
06:28
Yes, so why did we choose QGIS as the central component of the toolbox? Firstly, because QGIS has this processing toolbox, I don't know how many people are
06:40
familiar with processing toolbox in QGIS, but basically it allows you to call the algorithms from the other algorithm providers. So basically for most of the software there, the processing toolbox allowed us to access the algorithms straight away. So basically,
07:01
we didn't have to do any work. We could base our development on what was already there. And the global Africa toolbox is based on a previous similar software package which was called Voice Water Operation and Information System, I think. And that was developed at the
07:20
time of Quantum GIS 1.8. So back then, the processing toolbox was called sextant toolbox. But the idea was the same that you could access the algorithms, for example, from grass or from Orfeo toolbox within the QGIS GUI. Then QGIS also has this Python plugin framework, so it allowed us to extend the functionality if it was not available already
07:48
in QGIS. It is user-friendly and widely used, so we conducted many trainings in Africa and most of the users were already familiar with QGIS and used it before. So this is an environment
08:03
that they already know how to operate. And there's active development community, so the processing toolbox is becoming more and more robust, more and more stable, the user interface is becoming more and more user-friendly, and the software is continuously
08:24
improved. And last but not least, it's free and open source. If it wasn't, we would not be able to use QGIS in this project. Okay, so what did we do to develop the toolbox? As I said, a lot of the work was already done for us because of the sextant or later processing toolbox
08:43
which already exists in QGIS. But we had to fill in some gaps in the functionality, and we did it by developing QGIS plugins mostly. So there is a provider for algorithms coming
09:00
from the snap and beam software, there's the workloads provider, and we did some hydrologic modeling scripts as well. Then throughout the development of the toolbox, we were quite an active user of the processing toolbox and also looking for the code.
09:22
So whenever we found bugs or missing features with the pull requests into the QGIS project, we did some processing scripts and models to automatize certain tasks within the workflows. And we created an installer as well because there are all the different software components
09:43
so to make it easier for the users, there's one installer. So this is kind of the software development part, but probably an even larger part of the project was designing the workflows for those six applications and creating training materials. So basically, the workflows
10:00
capture the knowledge of the experts which were part of the project consortium, and then it is delivered to the users in those easy-to-follow step-by-step workflows. So here I'll focus mostly on the software part. So going through those three plugins,
10:22
they are designed as standard QGIS plugins, so they were kind of developed for the toolbox and are integral part of the toolbox, but they are quite generic, so they can be used as well outside of the toolbox. They are available in the QGIS plugin repository. So yeah, the first one is the algorithm provider for the snap and beam software.
10:42
So many tasks required to process Sentinel data cannot be performed easily by other software apart from snap, so that's why we import it. Only selected algorithms, not all of them, snap has many more, but the ones which are required, we created this algorithm provider,
11:04
so they are now available in the QGIS processing toolbox in the same way as grass algorithms or field toolbox algorithms or GDAL. The interface between snap and QGIS is this graph processing framework. So this is a
11:24
kind of command line interface of snap, but it's a bit more than that. It also allows the creation of graphs, so an example is there. Basically it's very similar to the idea of the QGIS modeler, except the graphs in snap, so in the QGIS modeler the output is saved to this
11:45
between any two algorithms. In the snap graphs, the input is read once and then only passed through memory for all the steps and only saved at the end. There are some other kind of enhancements there for processing large data sets, so for example automatic splitting of the
12:05
data set into tiles. We use this graph processing framework interface, which means that we can create graphs using the QGIS interface and then they can be run within snap and also graphs created within snap can be run through the QGIS processing toolbox.
12:27
The second plugin was the processing work provider. Basically it allows the creation of those step-by-step workflows. On the left here is the dialogue for the processing workflow
12:46
creator. You have the list of all the processing toolbox algorithms and models and scripts and they can be selected and put into the workflow. We have different tabs here for the
13:04
information panel, which can be styled using simple markdown syntax to give user instructions about what is happening in the step. Then when the workflow is saved, when it is run, we get a dialogue like this, so again the instructions on the left and then
13:22
the setting of the input and output parameters. When the current step is finished, the automatic step automatically appears. The workflows can be combined into workflow collections, so basically it's a directory holding the workflow files, which are just text files.
13:45
The directory also contains a collection configuration file and basically a collection has its own icon and has its own style. For example, here we can see that the headings are in green, so this is for the styling, so you can make each collection look unique.
14:04
And then all the workflows within a collection appear as a separate category in the processing toolbox, but as well there is this button on the toolbar, which allows you to display the collection dialogue, so the library of all the workflows within each collection.
14:21
It's easy to create new collections and new workflows, so it can be used for other purposes as well. And then the hydrological model, this is not really a plug-in, it's more of a collection of scripts, so it's fully Python-based and we also use Taldem for some DM processing tasks. It has QGIS processing GUI and it's a quite simple hydrological model
14:47
based on the Bodeco framework, so it's a rainfall runoff model and has a river routing component as well. It is designed for poorly gauged basins, so it's supposed to use not so much
15:01
in-situ observations, so all the data required to run the model should come from satellites, so it mostly depends on rainfall. And there's a publication there if you are more interested. This was done by our colleagues from DTU, which is the Danish Technical University. Okay, so we have those six product types and then we validated them at a number of wetland sites
15:24
in Africa to see that the workflows are fit for purpose. And now going quickly through the different product types, so first one is wetland inventory, so basically classification into wetland non-wetland. It uses high resolution data, so Sentinel-1, Sentinel-2, Landsat-8,
15:43
so both radar and optical, and it's based on time series analysis of those observations. Wetland habitat mapping, so more detailed classification within the wetland areas de-alienated in the previous product, again based on high resolution satellite data.
16:05
And using Landsat we can go back in time, and then for example here it's a wetland in Egypt and we can track the evolution of land cover classes associated with our anthropogenic land
16:20
use, for example aquaculture ponds and artificial surfaces, how they grow over time in the wetland area. Another one is inundation regimes, which tells us how often a given area was flooded in a specified time period, also the minimum and maximum surface water extent, so this tells us something about the wetland health. There's the water quality of open water bodies, so suspended
16:46
sediments, chlorophyll concentration, and floating vegetation. This is based on medium resolution data, mostly Sentinel-3 and MERIS, so 300 meter resolution, but as well it can be performed using Sentinel-2, so 10 meter resolution or 20 meter, and then we can monitor for example sources of
17:05
pollution within those water bodies. River basin hydrology, so simulating or modeling discharge at the given point in the river, important information for wetland managers, but as well can be used for things like flood forecasting and flood warnings. And finally mangrove mapping,
17:26
so again kind of classification workflow, specifically focused on mangroves. Okay, so in summary, yeah, the toolbox is free and open source, so based fully on open source software, it's easy to operate because of the wizard-based processing, it's user-friendly
17:44
because it uses QGIS as the main user interface. And just the last slide, what's going to happen next, so the toolbox is already, well a couple versions of the toolbox are already out, currently they are based on QGIS 2.18, but we got some funding to continue the development, so
18:02
now we are working or we start working very soon on moving everything to QGIS 3, so the plugins, the workflows, the scripts will be imported to QGIS 3. And then as well, one of the main issues that our users experience is lack of reliable internet connection and computing resources,
18:22
and when processing time series of Sentinel data, you usually need to download quite a lot of it and to have a good machine, so we are kind of building a cloud-enabled toolbox, so the users can still use the QGIS interface, but then when they want to execute the algorithm, it will happen on a server based on QGIS server, and then the final result will be downloaded by the
18:47
user, so they won't have to download the input data and they won't need to have the computing resources at home. Okay, thank you. Okay, anybody, does anybody have questions?
19:10
I'll bring the mic to you. Thanks, hello? Yes, so in the recent versions of QGIS,
19:29
the TAO dam was removed as a provider, and I was just wondering, does that affect your toolbox as the output and the package itself? It was removed as a provider coming by default with
19:42
the processing toolbox, but it can still be accessed as a plug-in, or if it's been removed as a plug-in, then we'll add it back as a plug-in, I'm not sure. Yeah, and sorry, it's just, I did not ask the question well, actually, because it was removed because of a license, change in license, I think, and it's open source, I think, but with a different license, and I was wondering whether that licensing would create an issue for you or not.
20:05
Okay, I have no idea, I mean, I didn't look at the licenses, but I think if they're all open source, we can still package them, I mean, they're kind of installed separately, so there's this one installer, but it kind of calls the installers of the different software, so the user has to
20:21
accept the license conditions of each software separately, so hopefully that would address this issue, but I guess we'll have to look in more details. Thank you, any other questions? Can you maybe tell us more about documentation for the project and the tool, where to find,
20:50
or what type, or if you have tutorial, or whatever? Yeah, they are tutorials, and I think they are available from the project website. If they are not, then there is a contact form on
21:01
the project website, and then we can distribute them, but I think they should be there. So they are the tutorials for the workflows, basically, and then the documentation of the actual software, well, the Git repositories are there, GitHub repositories, and yeah, I mean, the plugins themselves are not super complicated, so there's not so much to document
21:25
there. Any final question? Okay, let's thank the speaker again.