Red River Basin Decision Information Network
On-Line Workshop
September 18, 2002 - 12:00 PM CDT

Web-Based Flood Decision Support
for Red River Valley Flood Prediction and Management

Rick Bowering
Water Branch, Manitoba Conservation
and
Nathan Diehl
Manitoba Region, Prairie Farm Rehabilitation Administration

Amy Sebring, Moderator

[Introduction]

Amy Sebring: On behalf of my EIIP partner Avagene Moore, myself, and the RRBDIN, welcome to the Red River Room of the EIIP Virtual Forum. Our session today will be an overview of the new flood decision support system for Manitoba, the Red River Valley Flood Prediction and Management tool.

[Participant instructions omitted]

Now I am pleased to introduce our speakers:

Well-known to many of you, Rick Bowering has 32 years experience in the field of water management, both for flood control and water supply. As Manager of the Surface Water Management Section for the Province of Manitoba, he coordinates preparation of flood forecasts, streamflow monitoring flood control structure operation, flood damage reduction studies, water supply studies, and reservoir regulation studies.

Nathan Diehl has been working with PFRA Manitoba Region for the past two years as the primary Application Developer for the Red River Flood Decision Support Project. He began the project after completing studies at the University of Saskatchewan in Electrical Engineering and Computer Science. In addition to the Red River project, he is also working on other web-based GIS applications such as the web-site for the Manitoba Riparian Health Council.

First, Rick will provide us with some project background, and then Nathan will walk us through some of the features. And now I will turn the floor over to Rick to start us off.

[Presentation]

Rick Bowering: Thanks Amy. As you are aware, the Red River Valley is very flat, dropping only about one foot in five miles. The gently sloping nature of the landscape makes the valley prone to widespread overbank flooding.

There is a long history of flooding in the Red River Valley. Major flood events at Winnipeg have been observed in 1826, 1950, 1979, 1996, and 1997. Almost 2/3 of Manitoba's 1.1 million people are located in the Red River Valley. The 1997 Red River flood was the largest flood on the Red River in Manitoba in over 100 years. 2,500 km2 (1,000 sq. miles) of valley lands were inundated, creating a lake 80 km (50 miles) long and up to 40 km (25 miles) wide.

Management of the 1997 Red River flood in Manitoba was generally successful. The flood protection works for Winnipeg and the ring-diked communities in the valley all functioned well. The only community that was flooded was Ste. Agathe, which had no dike in place. However there was considerable damage in rural areas. Some 28,000 people were evacuated, 250 homes were destroyed, and a further 630 homes were flooded.

The experience indicated that flood fighters require better information. In particular they need more site-specific forecasts, and a more effective way to get that information to the valley residents.

Right after the flood Canada and Manitoba signed "The Canada-Manitoba Partnership Agreement on Red River Valley Flood Protection". Funding under the agreement totaled $130 million CAD funded equally between the federal & provincial governments. The bulk of the funding went towards flood-proofing individual homes and businesses and in building ring-dikes around more valley communities. However some funding was made available to develop the GIS Decision Support system that we are discussing today. Now I'll describe the problem we are addressing with the Website.

Up until 1997 our forecasting procedure gave forecasts at specific points along the main stem of the Red River. Normally these are the communities along the river. This has worked well for us when the river remains in the channel. But in 1997 flooding extended far over the flood plain. The river found new channels, which were largely constrained and defined by the road network.

We had a very difficult time providing site-specific forecasts for farmers who needed to know how many sand bags to place. In 1997 we stationed staff in satellite offices in the valley with maps showing maximum depths observed in 1979. They then compared the 1997 forecast for the nearest forecast point with the peak level recorded at that point in 1979 and then "guesstimated" what the peak level might be at the farmer's location in 1997. After the flood was over we knew that we needed two improvements:

(1) We needed a better forecasting model that could estimate flows and levels all over the flood plain, and

(2) we needed an easier system to provide that information to the residents of the valley.

With respect to the first need, a forecast model was developed for us during the IJC study and we are now using it operationally. It is the Mike-11 model.

The second need is being addressed by the Web-based decision support system.

When we started this project in May 2000 we first met with potential end users to find the type of system that would be most useful with them. We met with them twice more during the development of the system to show prototypes and to solicit further input. The feedback we received was that the system must be easy to use, accurate, and provide appropriate tools. They advised on tools that would be useful to them, as well as layers they would like to see included.

One critical component of the system was the development of an accurate Digital Elevation Model (DEM). This is a digital surface representation of the topography for a specified area of land. This was needed so that the system would be able to compute flood depths anywhere in the valley. During the IJC study a DEM was created for the area between Ste. Agathe and the floodway entrance using LIDAR (Laser Imaging Detecting And Ranging).

As part of the current study we extended the area covered by LIDAR south to the international boundary, and north of Winnipeg to Lake Winnipeg. This provided us with geospaced topography to a vertical accuracy of about six inches. The LIDAR information has proven to be very useful. The following slide shows a 3-D representation of the floodway entrance taken from the LIDAR data.

[Slide 1]

The developer of the decision support system was Nathan Diehl. He will now describe the system.

Nathan Diehl: Thanks Rick. I will start out on the live site, and then I will use some static screenshots to explain some of the features.

When you enter in the URL or click on a link to the site, you will be brought to the splash page. From the splash screen we can simply select "Enter" to get into the site.

http://geoapp.gov.mb.ca/website/rrvfp/

Within the main page there are many pages of information you can look at, including a form that will allow you to send comments back to us. If you click on either link to the map you will be brought into the application itself. These links are labeled "Map" on the Menu Bar and "view map" near the text. This will bring up a disclaimer. After you read the disclaimer that the lawyers worked so hard on, scroll to the bottom and click the "Accept" button to enter the site.

This will bring you to the application itself. As you can see on the following slide, the application is made up of several parts and I'll try and explain what each part does.

[Slide 2]

At the very center of the screen you will see the map itself. Above the map you will see a toolbar with tools for changing the view. You can use these to zoom in or out, or to pan to different areas of the map. You can also use the previous view button, which will undo the last view change you made.

To the left of the map are two small windows. The top one contains an overview of the map. It shows you southern Manitoba and a red square to indicate what is currently showing in the main map window. The lower one shows a legend that tells you what the markings in the main map window mean. When there are many things on the legend at once, you can scroll the window down to view the lower parts of it. To the right of the map are the more powerful tools. We'll go through them button by button.

Identify button is your way of getting information on any map feature. Simply click on this button and then click on the feature you are curious about. A window will pop up with information on any features that were where you clicked.

[Slide 3]

The Coordinates tool is a quick way of finding out the location of any place on the map. Click the coordinates button and then click on the map and a window will pop up with the GPS coordinates of that location in several formats.

[Slide 4]

The Water Levels button brings up a window with two tools in it.

[Slide 5]

From this window you can choose "View Water Levels" and then click on the map and a window will pop up with the predicted water levels in the area you clicked for the entire flood. Use this tool to get an idea of when and how high the water will come at any point in the valley.

[Slide 6]

The other option available from the Water Levels menu is labeled "View Extent of flood from a certain date" This allows you to have a graphic representation of the flood shown for any date you wish within the flood period. It is useful for finding what areas ought to be wet or dry and when they will be so. To clear the map of this graphic you can click the clear button. It is the last button on the toolbar to the right of the map.

[Slide 7]

The sand bag tools, both found under the "Sand Bags" button allow you to estimate the number of sandbags required to protect an area based on the flood model. The application automatically chooses the highest predicted flood level and uses that in it's calculation. For both tools you can select the amount of freeboard that you would like to use in the calculations.

[Slide 8]

The first of the two tools you can choose from the Sand Bag menu is helpful for determining the number of sandbags needed to protect a dike or pad in the valley. Simply click on this button, and then click on the dike or pad in question. Unfortunately we do not have information for all pads and dikes in the valley yet, but we will be adding more information as time goes on.

[Slide 9]

The second Sand Bag tool allows you to draw lines on the map where you would like to sandbag and the application will estimate how many sandbags you will need to protect this area based on the local land relief and the predicted maximum water level.

To use it, click on the button, and then begin clicking on the map where you wish to sandbag. You will notice that each click adds a point to the window, and also draws a green dot on the map. The green dots are automatically joined with a green line to show where the sandbags will go.

[Slide 10]

When you are done with the tool, simply click done in the Sand Bag window or right click on the map. You will be given a chance to choose the amount of freeboard you wish to use, and then you will be shown the results of the calculation.

[Slide 11]

Another helpful tool on the toolbar to the right of the map is the measure tool, which allows you to measure distances on the map. It works similar to the Freehand Sandbag tool -- you simply select the tool, and then click on the map. A red dot will be left at each point that you click and lines will be drawn to connect them. The distance from the last dot to your cursor is listed in the "From last point" box, and the total distance along the lines is listed in the "Total" box.

[Slide 12]

The Layers button on the menu to the right of the map brings up a window with two choices.

[Slide 13]

The first, "Choose Mike 11 flood model" let's you choose the flood model that the application is currently using. This is handy if you wish to do calculations on what happened in 1997 or what might happen if a particular type of flood were to occur.

[Slide 14]

During most of the year there is no current flood to worry about so you will see what you just saw, a 1997 model and a few others. During a flooding event we would add a 'current flood' layer with the most recent Mike 11 data.

The second button in the Layers menu allows you to choose what type of information you wish to see on your map. If too many features are cluttering your map you can turn some off here. There are also some very interesting features you can turn on here.

[Slide 15]

Please note that some features may not show up on the map even if you turn them on if you are zoomed out too far. Some of the more detailed datasets would take very long periods to draw if they were shown when you are zoomed out and would slow down the application for others. If a feature you want to see is not visible, try zooming in more.

The Find button brings up the biggest menu so far, with a total of 4 items you can select! You can use this tool to search for a certain location on the map and zoom directly to it. You can also use it to highlight certain features on the map. I'll just explain one of these, as they are very similar.

[Slide 16]

The quarter section lookup tool will find a land location for you. Enter in as much information as you know to limit the search.

[Slide 17]

If we enter NW quarter of section 12 and click search it will return a list for us to choose from and highlight them all in yellow. If we choose one, it will be highlighted in orange and the map will zoom to that location.

[Slide 18]

The Save View tool simply saves the current map extent with a name so that we can easily return to this area later. Just zoom to the extent you like, click Save View, type in a name and then click save. After this you can navigate back by clicking save view and then clicking on the view you wish to see.

[Audience Q & A ]

Amy Sebring: Thank you very much Nathan. We will move on now to our Q&A/comment portion.

[Participant instructions omitted]

Question:

Jayant Deo: The tools that you just showed are truly impressive. Could you explain Mike-11 model?

Rick Bowering: The Mike-11 model is a 1-D model like Floodwave. It was developed by the Danish Hydrological Institute. It's main attraction is that it has lots of post processing tools to understand the output. The only drawback I can see is that it is not free.

Question:

Mark Deutschman: Rick, how long will it take you to create a new shape file from the Mike 11 run during an actual flood event?

Rick Bowering: That is an excellent question Mark. A typical run takes about 30 minutes or so, but the time will be taken up in updating the data files and in fixing instability problems. It will be an interesting challenge when we get into our first real flood.

Question:

Chuck Fritz: How did you deal with liability issues? Were there questions or issues raised here?

Rick Bowering: Chuck, we discussed this with our legal advisors. The disclaimer is supposed to take care of it all. However there is likely still some exposure if we make an error, but that is no different than the info we gave out in 1997. Just a different tool.

Comment:

Paul Bourget: Comment really, you might want to consider linking to DHI's site for the MIKE-11 info within RRBDIN.

Bob Halliday: http://www.dhi.dk

Question:

Ted Bailey: Have you tested the model against real data for an area sandbagged or dyked in 1997 to get a feel for accuracy?

Rick Bowering: Yes Ted we have tested it quite extensively.

Ted Bailey: What did you find for sandbag estimates?

Rick Bowering: In particular, as part of our EIS process we ran the model to assess the impact of the proposed flood protection work after the flood. Generally it works very well, but only to the level of detail you put into it. The problem I see with sandbag estimates is the accuracy of the line you draw.

Question:

Brian Fischer: Rick, obviously the most important piece of information is the DEM, but how important is it to have a inventory of hydraulic structures to input into the model?

Rick Bowering: In Manitoba the main structure we have is the floodway control structure and quite a bit of work went into getting it right. We don't have any other operational structures that would have an impact, so the rest is things like culverts, roads, bridges.

Comment:

Jayant Deo: Just a comment: I think the liability issue can be explained by reminding people that (a) it is a tool for guidance, and (b) there are no guarantees.

Rick Bowering: Good comment.

Question:

Ted Bailey: Rick, Paul mentioned the RRBDIN. I wondered how this initiative relates to that one, which as I understand it is mainly U.S. focused at present.

Rick Bowering: The border remains a problem.

Ted Bailey: How do we get better coordination?

Rick Bowering: Our model starts at Grand Forks, but we only show data for Canada. So far we have not found a workable way to unify activities in the U.S. and Canada.

Question:

Mark Deutschman: Rick, are there any plans to set-up a Web Mapping Service that can be accessed via the RRBDIN with BasinViewer? I would really like to see this. This would help resolve the border issue.

Rick Bowering: I think that would be a good idea, but so far our concentration has been on getting this up and running.

Question:

Bob Halliday: Rick, what to you see as your short term and longer term next steps?

Rick Bowering: Our next step we are working on now is to extend the system to Lake Winnipeg. We have the map extent, but need to load more layers, and we are working on those. My next issue was referred to earlier. That is how to work this operationally during a flood event.

Question:

Jayant Deo: How difficult it is to incorporate changes in the structures like bridges, levees, etc. into this tool because they do make the difference to the flows?

Rick Bowering: Like any model, the input files need to be modified, and that is not too hard with Mike-11, but must be done. The bigger problem is getting the new configuration running in a stable manner. That is a problem shared by all dynamic models.

Comment:

Paul Bourget: We plan to migrate the RRBDIN to RRBC control that should start to move it to a basin-wide system, which was the plan for what it should be all along (point of clarification on Ted's earlier comment).

Question:

Mike Bender: How much did this project cost and do you have an idea of expected benefit/cost-effectiveness during a flood? (Somebody had to ask!)

Rick Bowering: I'm trying to look thru my files. It was something like $1 million. Regarding cost/benefit, hard to say, but I can say getting this kind of info out in 1997 was a major problem.

Question:

Chuck Fritz: What about outreach - placing these tools in the hands of, and educating, local residents and decision-makers? (Amen to Paul's RRBDIN comment.)

Rick Bowering: This is very important. We are working with the local government officials to educate them, and are discussing holding workshops this winter for valley residents to try it out.

Question:

Mark Deutschman: Rick, my understanding is that if your using ArcIMS that it isn't too difficult to set up a WMS server. Can I call you about this? I would like to add the Canadian data to the RRBDIN. We have done this successfully with EROS (USGS).

Rick Bowering: Sure call me, but I'll probably just forward your call to someone who understands it more than me.

Question:

Jayant Deo: I think one of the important aspects of outreach should be to emphasize the fact that people should not develop a false sense of security.

Rick Bowering: Excellent comment. In fact, in an ideal world we would probably not have people living in the flood zone!

[Closing]

Amy Sebring: Thanks very much to Rick and Nathan for taking the time to share this with us today, and thanks to all our participants. We will have the transcript and slides posted soon, and I will send a message out to the list as to when and where it is available. Our formal session is adjourned but please help me thank our presenters.