Minot changes flood model

RURAL ESTEVAN, Sask. -- It was a sunny day in August as Doug Johnson, a water resource engineer with the Saskatchewan Watershed Authority, drove on the gravel road that crosses the big earthen dam on the Souris River.

The reservoir that stretches some 35 miles behind the Rafferty Dam here on the Canadian Prairies is the biggest of the four dams in the Souris Basin. It's the linchpin of a basin-wide system that provides flood protection for Minot, North Dakota's fourth-largest city some 100 miles away on the other side of the border, and scores of smaller towns, including parts of Estevan, two miles east of the dam.

Johnson, who oversees the watershed authority's dams in this region, was explaining how the system works and how it really had been working two months prior when many of those towns flooded.

Engineers had designed the system with spring flooding in mind. Their model, based on historic weather and hydrological data, showed that snowmelt is the most common cause of flooding here.

Summer rains, never easy to predict very far in advance anyway, had not been much of a flood threat in this region. In fact, it usually goes dry and there's not enough water.

The watershed authority had followed the operations manual and lowered reservoirs based on snowmelt and some predicted rain. But the rain that did come in June was more than anything then in the record books. A freak occurence.

"We've done calculations and we think through the rainfall event we could've refilled Rafferty full to empty two to three times," Johnson said, sounding slightly amazed still. "That's how much water came down."

He stopped his vehicle on the side of a gravel road down in the flood plain beneath the dam and pointed to a house with a tell-tale mud line along the sides. The water must've been at the first floor windows.

The dam, Johnson said, was supposed to protect homes in this area from a one-in-500-year flood.

U.S. officials, including North Dakota Gov. Jack Dalrymple, are now talking with Canadians about changing the operational plan for the dams that both sides agreed to in 1989 to account for torrential rain. They also propose raising Lake Darling Dam, the one dam that's on the U.S. side, and building a system of dikes and diversion in the Minot area similar to the one built in Grand Forks after the 1997 flood.

The 2011 flood, which forced the evacuation of some 11,000 in Minot alone, had changed the historic model.

Tools at hand

But those changes are in the future and each face their own difficulties.

What's available now is what was available last summer: Three dams in Canada, each astride a different river, and one in North Dakota.

Rafferty Dam lies on the Souris and has a reservoir that, at maximum capacity, is nearly one-and-a-half times bigger than the reservoirs of the other three dams combined.

Boundary Dam, with the smallest reservoir, lies on Long Creek a tributary of the Souris. Forty miles east and downstream, Alameda Dam lies on Moose Mountain Creek, another Souris tributary.

Lake Darling is furthest downstream, just north of Minot.

The 1989 agreement says that the Rafferty reservoir must reserve 265,000 acre feet for flood storage, enough to cover an area 28 times the size of Minot in a foot of water. It's about half of the reservoir's maximum capacity.

The agreement also says 113,000 acre feet, or three-quarters of the maximum capacity of the Alameda reservoir, must be for flood storage. It says that because U.S. taxpayers paid about a third the cost of the dams.

The agreement also includes an operations manual as thick as a dictionary that determines the water level in the reservoirs based on anticipated snowmelt. The more snowmelt, the lower the reservoirs should be at the start because it's anticipated they would refill quickly.

Johnson said that's relatively easy to figure out because all the water's already on the ground.

But adjusting for rain is much harder because it can't be predicted very far in advance and even as it's happening, the available rain gauges and radars in Canada aren't enough to get a very good picture of how much rain is falling.

If there is criticism of the flood fight from some U.S. officials in 2011, especially those not involved in dam operations, it was that the Canadians didn't alert anyone in time about all the rain that was headed downstream. Minot had but days to raise emergency levees in response.

Johnson said dam operators here didn't get the news a whole lot earlier. One day the predicted rainfall was 3 inches and the next it was 7, he said.

Changing tactics

This past week, Gov. Dalrymple and Sen. John Hoeven, R-N.D., said they met with Saskatchewan Premier Brad Wall to talk about amending the 1989 agreement so that rainfall would be accounted for, and the premier agreed with them.

Hoeven said in a statement that the goal was "to ensure ample storage in the reservoirs for both spring melt and summer rains." Technical experts on both sides will figure out how an amended agreement would address that goal, he said.

In some ways, the flood fight in 2011 did account for rain. Officials on both sides of the border indicated that dam operations were more flexible than the 1989 agreement might suggest.

That's because U.S. and Canadian officials on the International Souris River Board, which implements the agreement, decided that it was necessary to deviate from the agreement.

It helped that people with direct control of dam operations such as Johnson and Col. Michael J. Price, the commander of the U.S. Army Corps of Engineers St. Paul District, are board members.

Price said in July that board members talked frequently and, during the flood fight, often arranged formal meetings by phone within hours to give dam operators the green light to go outside the operations manual.

It may take some time to renegotiate the 1989 agreement, he said, because there are a lot of agencies involved.

In the meantime, the board's Canadian co-chairman, Russell Boals, said in September that there seemed no reason why the board couldn't deviate from the agreement again in the next flood fight.

But there is a limit to any change in the agreement.

Flood control is important for Canadians, but they didn't build the dams entirely for that purpose. The dream in southern Saskatchewan was to ensure a stable water supply during dry cycles and droughts. It was the United States that paid for flood control after the U.S. Army Corps of Engineer's proposed dam in Burlington, N.D., north of Minot was opposed by landowners and their allies.

For dam operators, then, there will always a tension between leaving just enough room for flood storage and leaving too much room.

"What you're constantly thinking about when you're trying to manage a water supply storage reservoir is 'I gotta get it full because, next year, I might not get any runoff. I might not get any run off for a couple, three years and then my water level's going to be way down,'" Johnson said.

He added that there is more wiggle room today because there isn't a lot of demand yet for the water in the reservoirs. But dam operations become less flexible as farms invest in irrigation or a big ethanol plant comes to town, he said.

In North Dakota, the governor is seeking more flexibility by pushing for an upgrade to Lake Darling Dam to store more water. He said last month in a statement after a meeting in Minot that both U.S. Fish and Wildlife, which owns the dam, and the U.S. Army Corps of Engineers, which controls it during flood fights, are open to the idea.

Second defense

Still, these would be but incremental improvements compared to the system of dikes and diversions under development as a second line of defense.

Tim Fay, a State Water Commission official who heads the project, said it's being designed with the assumption that there would be no changes to the 1989 agreement. For one, he said, change is difficult especially when it involves two national governments and it can't be counted on.

"The other thing is, changes that could be made, they're not real extensive," he said. The changes might bring some "smaller scale" improvement, he said, but they wouldn't have made a difference this past summer.

The amount of water dam operators had to contend with was simply "off the charts," he said.

Unlike Grand Forks' and East Grand Forks' dike system, which protects just the cities, the Souris River dike system would protect the entire basin on the North Dakota side of the border, not just Minot.

The first phase begins 45 miles northwest of Minot at an area called Mouse River Park, follows the river through the city and goes past Velva, 20 miles to the southeast.

Most of the small towns and residential developments along the Souris would be protected by relatively small sections of dikes anchored against the sides of the river valley.

It's inside of Minot where the dikes and diversions get complicated. The alignment that the Minot City Council approved this past week includes dikes all long most of the river's path and two diversions channels as wide as 480 feet that would straighten the bends. The point of that, Fay said, is to reduce the number of homes that need to be bought out and minimize costly dike construction.

The next phase would include other communities hurt by flooding.

Downstream areas such as Towner didn't get much attention last summer, Fay said, but they too had some property damage and a lot of economic damage when farmers couldn't plant and ranchers couldn't get trapped livestock out.

The preliminary design of the first phase is scheduled to be completed by the end of the month.

As engineers push toward deadline, they'll be guided by historical data the same way the dam engineers were. The dikes would be high enough and the diversions wide enough to fend off the 2011 flood though with three feet tacked on, just in case.

But as big as the last recorded flood was, there's no telling when a bigger one will come along.

Scientists like Steve Buan, a National Weather Service flood forecaster who covers the Souris Basin, like to point out that just because it hasn't happened yet doesn't mean it won't in our lifetime. "When I go out and talk to public, I say, 'No, no, things could be twice as bad as what they were in the big one because it's happened. And I list off all these events where things were twice as bad from a streamflow standpoint."

Grand Forks in 1997 and Cedar Rapids, Iowa, in 2008 came to mind for him.

Suppose Minot's 2011 flood was a 500-year event as some have speculated, statistically, it has a 0.2 percent additional chance of occurring with every year that passes. Given that there has not been a flood of that magnitude for at least 108 years -- that's how far back the river flow records in Minot go -- the probability of a 500-year flood last summer should've been around 19 percent last summer.

And a 1,000-year event? It'd be 10 percent.

Reach Tran at (701) 780-1248; (800) 477-6572, ext. 248; or send e-mail to ttran@gfherald.com .