Adopting OGC Standards in a Flood Alert System
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 | 188 | |
| Author | ||
| 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 | 10.5446/31589 (DOI) | |
| Publisher | ||
| Release Date | ||
| Language | ||
| Producer | ||
| Production Year | 2014 | |
| Production Place | Portland, Oregon, United States of America |
Content Metadata
| Subject Area | ||
| Genre | ||
| Abstract |
| |
| Keywords |
FOSS4G 2014 Portland180 / 188
7
10
14
15
16
23
24
25
28
29
33
37
39
40
43
45
46
48
50
56
64
65
69
72
74
82
89
91
98
102
107
111
114
118
128
131
132
135
138
141
143
147
149
150
157
158
161
164
165
166
173
174
175
179
185
00:00
Event horizonScaling (geometry)Variety (linguistics)AreaInformationGroup actionMusical ensembleProcess (computing)State of matterBitLengthFlow separationLecture/Conference
02:38
Graph coloringRing (mathematics)Workstation <Musikinstrument>Reverse engineering
03:05
InformationAreaMassGraph coloringScaling (geometry)Service (economics)CuboidData structureRow (database)Constructor (object-oriented programming)Communications protocolDatabaseServer (computing)Execution unitLogical constantProcess (computing)Computer fileJava appletGoodness of fitMultiplication signComputer-assisted translationTexture mappingBound stateMathematicsQuicksortWater vaporFood energyRegular graphHand fanDistribution (mathematics)Workstation <Musikinstrument>Medical imagingWebsiteCross section (physics)Slide ruleFile formatGoogolGroup actionLibrary catalogCASE <Informatik>Web browserConfiguration spaceComputer configurationRight angleComputer programmingSubject indexingSatelliteArrow of timeQuery languageCellular automatonDirection (geometry)MappingSystem programmingProduct (business)Electronic visual displayPoint (geometry)Filter <Stochastik>Military baseLine (geometry)Vulnerability (computing)Shape (magazine)Active contour modelVirtual machineState of matterLocal ringSound effectCivil engineeringFilm editingCartesian coordinate systemCache (computing)Remote procedure call
12:48
DivisorPoint (geometry)Computer fileComputer configurationCodeMultiplication signDirectory serviceVector spaceServer (computing)Shape (magazine)DatabaseLocal ringImplementationSource codeInterface (computing)Link (knot theory)Kernel (computing)TelecommunicationSocial classService (economics)ComputerVirtual machineFactory (trading post)Maxima and minimaContent (media)Goodness of fitFile formatConvex setAbstractionBit rateSystem programmingInformationRing (mathematics)Graph coloringEndliche ModelltheoriePresentation of a groupMassMedical imagingCellular automatonMatrix (mathematics)Web browserData storage deviceData conversionSI-EinheitenType theoryState observerMathematicsExistential quantificationEngineering physicsForm (programming)Different (Kate Ryan album)Projective planeMathematical analysisResultantProduct (business)Semiconductor memoryMultiplicationQuantum stateGeometryMetreComputer programmingDirection (geometry)Raster graphicsFrame problemModal logicTexture mappingSoftware testingInternet service providerCoroutineAreaPhotographic mosaicComputer animation
22:31
Model theoryDecision theoryWorkstation <Musikinstrument>Medical imagingAreaMaxima and minimaProjective planeReal-time operating systemMatrix (mathematics)InformationStudent's t-testStreaming mediaPower (physics)DiagramProgram flowchartComputer animation
Transcript: English(auto-generated)
00:02
And Sao Paulo, just to contextualize, Sao Paulo is one of the state of Brazil, the biggest, not the biggest, but the most, with the biggest population of Brazil. And our work is related to mainly in the metropolitan area of Sao Paulo,
00:22
that you can see that the population there, it's very huge, about 20 million people living in this area, so very populated and with a lot of problems with the flooding, which is our main concern in this area.
00:42
Just to contextualize in which situation we use the GIS processing, I'd like to talk a little about Sao Paulo, the problems that we have during the summer. This is a typical day in the afternoon in Sao Paulo without rain.
01:01
And this is a bad day in Sao Paulo. In the afternoon, after a severe event of rain. So the same situation, without rain, with rain. It doesn't happen every day,
01:21
but sometimes it happens and it causes a lot of problems because imagine a city, Sao Paulo has 12 million people. So the traffic, normally it's a big problem during the harsh hours. But with this situation, it becomes even worse.
01:48
So Sao Paulo, we have some tools to deal with this situation. We use weather radar to do now casting
02:03
and for some actions to be taken in advance. So we have, nowadays we have S-band radar with a very nice accuracy. So this is the most important tool
02:23
that we have to deal with this problem. So this is the typical information of radar. There's on the left hand side, we have the scale. The hotter the color, the more intense the rain
02:40
and the lighter the color, the lower intense the rain. We also use some ground stations. We use to measure the level of the rivers and the intensity of the rain using some equipments installed in the weather stations.
03:04
And for doing this in metropolitan area, we have about 300 stations. We use GPRS to transmit the data or in some cases, a satellite.
03:23
You can see some charts here. The first one is the accumulated rain in time and the second one is the level of the river. This is good to, it would be good to realize that we have some levels, warning levels
03:44
catalogued previously, like attention alert, emergency and flooding. So in each case, the water reach this level, some actions must be taken by civil defense and governmental institutions.
04:02
This is just a river cross section sample in the river in a regular situation. This is river in a hard situation near the flooding level. Our main customers are the governmental agencies
04:20
like water and energy users and the civil defense, some civil construction and the traffic engineering also, they use this kind of information. Sao Paulo here is divided in many sub-areas.
04:46
And we do this kind of monitoring for each sub-area. So we have the color, a scale of color which shows that when it's yellow, it's in attention
05:01
and red, it's in alert. So at this point, I will pass for my colleague Ivan to continue and to show how we use the GIS tools.
05:20
Hello, good morning. In this map, you can see that we get this map from GeoServer and we have some static information and some dynamic information. The states are feed to the GeoServer to change the color depending of the status of the area.
05:46
Our infrastructure is based on virtual machines, locally virtual machines on VMware and remote virtual machines on Amazon. We use the Linux with basic CentOS with Varnish
06:01
for getting performance and cache because we have a lot of traffic, mainly in that site you saw in previous slide. That site, we have one day with more than two million hits in one day.
06:20
And when the information is created online, it spends a lot of processing and we now use each processing gets cached by one minute. Then we have some delay on the update, but just one minute, that's not too much, but gives a good performance with lower machines.
06:44
We use Java with Apache Tomcat 7 and GeoServer and our main database is the DB2. We have also Postgres and Postgres, but for smaller things. Our current database for telemetrical data
07:01
have 260 million records. It's almost 47 gigabytes of data. The structure is based on Varnish in front of our servers, all of servers, and then our request passes on Varnish to get cache and gives good performance.
07:22
All the services are virtual and they are redundant. In this slide, you only see the one box for each service, but they are redundant. What we use about the GIS services and the open protocols.
07:42
We use WMS, WFX for local data and external data. We got some data from outside also. We create on GeoServer and GeoTools images for publishing directly and to use internally
08:00
based on shape files and our products. We use also image maps generated on GeoServer to create hints, show hints about the places where the user puts the mouse over. Here we have an example about how we use GeoServer
08:22
to simplify our work. We have a product that show contour of the rain and it's the only one product generated by the radar. We have that as one shape on the GeoServer and we have some filters to show it
08:40
as many different layers for the user. And also we use CSS and SLD for defining the structure, the display. And for that product, I will show the imaging.
09:00
That shows that we have a contour on the rain cells and a small point in the middle of the cells. That is the case that I show on the XML before. On here, we have all the cells from the past. You can see on the right, the checkbox on CTR passado.
09:26
That one. And we index slides that we are changing. The current selected now. We have the forecast and here we have all them together.
09:44
We have this option that's very easy for the user to select and very easy for us to program because we just need to create an XML to configure our servers,
10:00
our application. We generate using the GeoServer many formats and also some formats are get directed to the user and use it on the browser and some formats are processed again by our application. In mainly, we provide PNG, GIF, JPG for the web browsers
10:25
together with the image maps. And for use on Google Word, we generate directly KML and PNG. Some images we do with transparency,
10:41
some images without transparency to make easier depending on where it will be shown to decide what's better. And in GIS, we use the two services, WMS, WFS. Here we have one of the changes on our system.
11:05
We have the map on the left that shows old system that we have a CAD, AutoCAD based map for the region. Every time we have to make a cut or change something, we have to do manually everything,
11:22
check if the bounds are correct. Everything would be made manually. We have a lot of problems with this, some errors, time consuming for updating and not good. Now, we have some maps that are directly from shape files
11:45
and some data comes directly from database that we can put together. Easy way, if you need to change the bounds, the cuts, we just have to re-execute the query. No problem to update and it's faster,
12:02
easy to decide what you want to over each layer, then it's very efficient. Here we have another example that was very difficult to implement in our old system. Now we have some arrows to show the wind direction,
12:24
directly over the rain. The colors are easy to update if you decide sometimes we need to change the color. Just one line of configuration and we can change everything.
12:40
Just for information, in this new system also, we have option to click in these colors and it will blink the same cells in image that are with same color and you can say, oh, where is 10 millimeter raining and that area will blink.
13:02
We have option to use external data like OpenStreetMap. Sometimes the user wants to see at street level. We don't have to create in our system, we just connect to OpenStreetMap to use this.
13:20
It's written without rain. More technical information. I don't know if all of you are technical or not. We have some things that we have do that take a lot of work to understand because documentation was not very good
13:41
but we have implemented our own data store to plug in the gel tools and that data store converts our data that comes from the radar. That's proprietary format from the company that made the radar but they provide the data formats and we created the data store
14:02
to integrate directly on the geo server. Currently, we implemented using the abstract file data store but Jordi said that it's not a good thing because it will be deprecated and I restarted to migrate to content.data store.
14:22
It's very easy to migrate. I have done during the workshop yesterday morning because all the difficulties was done before. For everyone that wants to make something like this, I recommend to use these two links of the documentation that are the data store
14:41
and also the links which tutorial have the sample how to create a CSV data store from the old method and new method. It's very easy to do this. For quick steps, you have to implement a factory
15:02
extends the data store and you have to put in a file the name of your class. That's the quick steps to create your own data store. For creating the data store, you have also to create the data source but it's on the tutorial you can see.
15:24
We have two basic implementations. One that reads our database and provides the layers, definition of the layers that we use and also we have the file data store that reads the file and show the vector data
15:41
for the geo tools, geo server to be presented and for doing this, it's not easy to being restart in geo server. I recommend to use J frame map for testing. You code in geo tools, show in J frame
16:00
and if everything works, you probably will have no problem putting it to working on geo server. You have just to put the jars in that directory. Data conversion. We have tried to use raster
16:21
and also we have tested with the vectors and we decided to use vectors creating a square cell for each point that from our radar data that becomes easier to style and show to the users.
16:41
Then creates a lot of memory use but our data is just a square. We don't have much problems. We have one of things that are difficult that we have some products that comes from the radar that one single file have multiple types of products.
17:04
Different shapes, different data. Then we created a virtual layer that we read the same file and provides a different schema for that data. We use the WGS84
17:22
that is EPSG4226 and geo tools does the conversion when necessary. We don't have to mind about the conversion problems, projection problems. What we want to do?
17:41
It's difficult to say we will do because some things changes and we have priorities. I expect to implement the SOS interface, the sensor observation interface and also implement the data store
18:00
with time support correctly. I have a simple implementation but I think it's not correct. We have to rewrite it. Offer direct WMS services for the external users. Currently, only our internal system uses the WMS
18:23
and also we are internal users but not for the public because it needs too many machine computing and we are checking what's the best way to provide this but we want to provide this kind of service for users
18:42
because the user will have a better interface, can use other programs, local programs instead of just web browsers. Multidimensional. We have seen a workshop that I understand now
19:01
that image mosaic can be used together also for vector and appear to be a good way to integrate with our system and next time I will get the code, I will try to implement that in our system.
19:23
I think it's an overview what we do and maybe you have some questions that can be used on the time we have here.
19:46
No, when you have the code, you have for the data, you have basically two options. We use raster, just points and when I call vector is like the shape files
20:00
but vector is internal format. All shape files will be managed by vector. As I'm converting my matrix to internal, I use the vector like the format.
20:24
You say how we go from this to that alert. That point is manually built on because we have many information and we have some problems
20:41
when we tried with some automatic informations because sometimes we get a sensor that gets a problem. It goes very high, too fast and not real. Sometimes it's real. Then the people that works 24 by seven
21:01
in the emergent system check the informations, get some of our internal alerts and really decides it goes yellow, it goes red based on some informations that the system provide but that image go to public and it's updated one minute by one minute
21:21
but just when one person select the information, it goes to public. We have many things that provide the information and one user decides this.
21:42
We have the radar uses square points. We have created a routine that make a simple conversion that defines the central point of the radar and we goes one kilometer, one kilometer
22:03
and create the square grids. We have tried with different routines and find one that gets better results. We have plugged grids models and decided by one.
22:23
That's different. We have, let me show what is not really on the presentations after presentation just for reference. We have that square and we have that one that you see that's not really square in the same area.
22:40
We can choose between the projections and decide what's better. We have decided for this square.
23:08
We also, we don't use next red. Yes, we use hydrological model and now we are using SWIM model from EPA
23:23
and we assimilate the data from weather radar and weather stations and this model, it runs in real time every 10 minutes. It makes a run to do a forecasting for this forecasting generates a flooding area
23:44
for the next maximum two hours. So we use the images from weather radar, images, the current imaging but also the forecasting generated by weather radar
24:02
which can give us one hour in advance with a good reliability. So this information is assimilated by our hydrological model. More questions? Thank you.