The Telltale Sand Boil; Or how Earthquakes led to a discovery of the Cause of Floods

The geographic setting of levees makes them vulnerable to earthquake failure by foundation liquefaction. Interestingly, the mechanics of seismic pore pressure migration is mirrored in floods. The case is made here that the view of levee failure by piping may be incorrect for many cases of failure. Accurate eyewitness observaation may be the key to correct diagnosis of levee failure.

0.5:aneqlev5.gif:674

We are examining here some levee failures that occur along the dangerous blue-tinted "levee failure belt" in California and elsewhere.

1:mfa1.gif:611

We focus on three deadly failures north of Sacramento near the Gold Rush city of Marysville

2:mfa2.gif:612

Thirty years after the famous "Christmas Eve" failure that flooded Yuba City, a similar failure happened in a similar geographic area on the other side of the river.

3:mfa3.gif:613

Photo of the 1986 Linda Levee after failure.

4:mfa4.jpg:614

None of the failures we are discussing involved overtopping of the levee. In fact, the history of levee failures in the U.S. indicates 50-75 percent of failures occur when the river is below the crest of the levee. Eyewitnesses agreed on a sequence of events like this.

5:mfa5.gif:615

How exactly do they fail? One traditionally conceived failure made is embankment seepage occasioned by seepage through the levee itself, aggravated by those great scapegoats of levee failure, ground squirrels.

6:mfa6.gif:616

But for decades levee failures have been associated with sand boils which do not appear to be related to the levee itself.

7:mfa7.gif:617

Studies by the Corps of Engineers on the Mississippi indentified a process of failure that involved underseepage as a fundamental start of the failure process.

8:mfa8.gif:618

It wasn't the first time I had pondered the destruction of levees. In the mid sixties as young engineer I had built a whole system of them in Thailand. They looked very handsome in the dry season.

8.1:img0041s.jpg:663

Here is the village of Takhop where we lived-- in the dry season.

8.15:img0035.gif:671

The rainy season was a different story; many of my levees failed inundating our village. Fortunately the people were quite used to these sorts of problems there, but I was left with a problem of how to do the job better.

8.2:img0046s.jpg:668

One day there was a festival and I was taking a photo of my daughter standing on one of the levees

8.3:img0033s.jpg:664

Suddenly down the road a line of elephants appeared, giving me a possible answer to the levee failure problem.

8.4:img0040.gif:672

We did appropriate tests of elephant soil compaction capability (ultimately featured in the new Yorker Magazine as a kind of joke). We wrote a paper on the matter. (Click here for more on this).

8.5:Img0001.gif:673

Meanwhile I found that studies by the California's Department of Water Resources in yearss past had identified an entirely different problem -- seepage that killed orchard crops after floods. It seemed to identify exactly the same condition that occurred in floods.

9:mfa8a.gif:619

Another clue was the apparent delay in the Linda failure -- nearly a day a year the flood peak had passed.

10:mfa9.gif:620

These ideas led me to conclude that the failure was a foundation failure -- brought about by deep foundation seepage followed by piping out of a large enough cavity to permit settlement of the levee.

11:mfa10.gif:621

During the several days of trial questioning I commented that another area that looked ripe for failure was down river a few miles at Arboga.

12:mfa11.gif:622

This prediction proved to be true. The day after New Years 1997 the levee broke at that very point.

13:mfa12.jpg:623

.causing the usual toll of death, property damage, and misery.

14:mfa12a.jpg:624

The break occurred just where history.

15:mfa13.jpg:625

.and geology suggested it should. See http://www.stanford.edu/~meehan/flood/aeg.html for more on this.

16:mfa15.gif:626

Leading to fame for a day.

17:mfa14.gif:627

But I was not satified that I understood just what had happened in the fialure process. Here we had more eye witness history and testimony, featuring extensive boils.

18:mfa15.gif:628

Some quite tiny.

19:mfa16.gif:629

Others evolving into sinkholes

20:mfa17.gif:630

Visible from the air.

21:mfa18.gif:631

and on the ground.

22:mfa19.jpg:632

Interestingly similar patterns of boils were known to develop in earthquakes as here in the Pajaro River during the Loma Prieta quake.

23:mfa20.gif:633

They also suggested future levee failures -- as shown at Pajaro. (Note the correspondence of the failure with the intersection of the county line with the levee; this interesting coincidence is explored in http://www.Stanford.edu/~meehan/parajo/index.html)

24:mfa21.gif:634

Many people laughed at the claims by flood victims that just before the failure snakes come out of the ground. People thought "Go tell that to your bible class."

25:snake.gif:646

The comparison to Pajaro raised some interesting thoughts. Thinking laterally for a moment, we note that in earthquakes boils = liquefaction of soil = buildings tipping in Japan.

27:mfa22.jpg:635

.in Taipei, and at almost any other earthquake affecting an urban area in a lower river valley.

28:mfa23.jpg:636

Combining all these ideas suggests the following sequence: rising water and foundation recharge.

29:mfa24.gif:637

.leading to boils.

30:mfa25.gif:638

.and finally to foundation bearing capacity failure.

31:mfa26.gif:639

The traditional standard flownet suggesting low uplift pressures at the toe

31.1:stfn.gif:665

Bearing capacity failures of levees following earthquakes provide important insights

31.2:eqbf.gif:666

.leading to a new model of flood failure.

31.3:newmod.gif:667

Concentration of upward seepage beneath the toe of a sound, tight levee is the most dangerous.

32:mfa27.gif:640

Computer modeling shows that complete liquefaction occurs with upward gradients of less than 0.5 toe failure from loss of bearing capacity happens at even lower gradients.

33:mfa28.gif:641

As suggested previously the pressure in the gravel driving the upward gradient will eventually reach somewhere between 50 and 100 percent of the river level, the latter where the river accesses the gravel in its natural channel or through artificial pits, which have been long and wisely banned.

34:mfa29.gif:642

It was clear that the reptiles were able to foretell the failure.

34.5:snake1.gif:647

Except where rich quarry owners are able to change the rules. Or in the case of the 1997 failure where wetland restorationism prevails over engineering principles.

35:mfa30.gif:643

Toe liquefaction under 18 ft uplifet, 30 ft non-stratified topstratum. See flowneta.wk4

36:mfa31.gif:644

Pore pressures rise and fall beneath the toe of the levee

animation/enlargement

37:pp.gif:645

The Osaka earthquake destroyed many of the levees that protect the city. Fortunately for the residnets the river level was low and there was no catastrophic flooding.

38:osakalev.gif:648

State studies after the 1997 flood called for levee repair schemes that seemed unlikely to solve the problem. Certainly the type of slurry wall shown here would do nothing to prevent the type of failure that happened at Marysville.

39:slurry.jpg:649