Forecasting the Future of the Mosul Dam

Posted on Posted in GJ Open, In the Media, Superforecasting

by Welton Chang

Imagine getting an alert with the following headline: “Mosul Dam breaches; hundreds of thousands missing or dead.” The Mosul Dam lies to the northwest of the third largest city in Iraq, a country where U.S. forces have been engaged in some capacity since the invasion in 2003. Most people have never heard of the Mosul Dam and public interest in the potential for catastrophic collapse has been low (see Google Trends).

Red = Kim Kardashian, Yellow = Kanye West, Green = Syrian Civil War, and the non-existent blue line is the Mosul Dam. Source: Google Trends

News coverage has also been sparse, save for a New York Times piece in January of last year and a few pieces that covered U.S. Ambassador to the UN Samantha Power’s public warnings in March. Most recently, Dexter Filkins published a piece in the New Yorker this month. If the dam were to collapse, impact estimates put the number of immediate deaths at around half a million and the water from the Tigris would put much of Mosul under as much as 15 meters of water. Without adequate warning and evacuation, the total number of dead and displaced could exceed 1.5 million, the functional equivalent of every single road-related death in the entire world for an entire year, except it would happen in a matter of hours.


Probability of Collapse

Right now, on Good Judgment Open, more than 230 forecasters put the probability of this event occurring in 2017 at around 6%. That means that if 2017 was simulated and played 100 times, the dam would collapse in six of those runs. But simulated worlds and counterfactuals being what they are (we live in only one), it may be more useful to put the risk of collapse in terms of odds. A 6% chance translates roughly to 9:1 odds, making the chances of a dam collapse the equivalent of basketball star Stephen Curry missing a free throw, or to drawing a jack from a full deck of cards. If you live in the United States, you are ten times less likely to die in a car accident in your lifetime. The forecasters on GJ Open put the odds of a dam collapse as roughly the same as North Korea not conducting a nuclear test in 2017.


The Outside View: Base Rate of Dam Collapses

An avid reader of this blog might ask: “How many dams usually collapse each year? And how many of those dams are of a type similar to the Mosul Dam?” True, the forecasted probability on GJ Open can be miscalibrated (since even if the dam doesn’t collapse in 2017, a 6% chance means no collapse should happen 94 out of 100 times). U.S. Government experts who have examined the dam assess it could collapse at any time, with little to no warning, and this risk will grow during the high water season in the spring. The true probability of a dam collapse is difficult to know, if not unknowable. One alternative estimate by the Iraqi Minister of Water Resources puts the chances of a collapse at closer to 0.1%, although based on the language used by experts such as the U.S. Army Corps of Engineers, the real number is almost certainly higher. While dam collapses in general are more frequent than most probably realize (there have been 82 since 1900), the only comparably catastrophic dam collapse in the last 117 years is that of the Banqiao and Shimantan Dams in China in 1975 which killed 171,000 people. Today, there are approximately 57,000 dams around the world and close to 300 dams similar in size to the Mosul Dam. So the probability of any one of them collapsing is very low. The low collapse rate has to do with solid construction, regular maintenance and with the fact that most dams aren’t located in a war zone.     


The Inside View: What Makes the Mosul Dam More Dangerous?

Several problems specific to the Mosul Dam should inform and likely boost the estimate of a collapse over the overall rate. Because the Mosul Dam was built on a soft and ultimately unstable foundation, keeping it intact requires a careful and laborious grouting process, with technicians injecting tons of cement into large voids that regularly emerge inside the dam. This maintenance process is prone to interruption (like when ISIL briefly captured the Mosul Dam in August 2014).  Even the efforts of the Trevi Group (an Italian company contracted to repair the dam) provide only a temporary fix — the only real way to prevent dam collapse is to decommission the Mosul Dam and build a new one at a place in the river with a solid foundation. Ultimately, a dam collapse is by no means impossible and it would be foolhardy to assume that the probability of a collapse is zero.

A Worse Problem Than ISIL?

So how bad of a problem is a dam failure? For the sake of argument, let’s assume the real probability is around 6%. This translates into an expected cost of around 40,000 lives, more than the total number of U.S. casualties during the Korean War. Even the aforementioned higher estimate of 1.5 million casualties is a lower bound estimate: a dam collapse would impact all of the Iraqi cities along the Tigris, including Baghdad. This would put nearly 7 million people at risk. The same number of people live in the state of Arizona. Standing water would bring diseases, corpses, and washed-up bombs into major urban areas, making cleanup a nightmare. International rescue efforts would be complicated enormously or even precluded in some areas, by ISIL’s continued presence.

The scale of a Mosul Dam collapse has been termed “epic”, “staggering”, and “Biblical”, and even these words don’t do justice to the potential for harm. Despite this, the dam still has not breached U.S. public consciousness. Part of the explanation might be psychological. It is well-documented that people do not fully understand the scale of events, in effect ignoring the zeroes in a number like 500,000 potential dead. We also tend to naturally discount the probability of terrible events, because for the most part, they don’t happen. People also process spatially and temporally distant objects different from those right in front of their noses — a dam collapse in far away Iraq, a place already plagued by awful events, may seem like an abstraction to most Americans. And there are many potential explanations for why an event like the Red Sox winning the World Series seems more plausible than a Mosul Dam collapse, even though the two have similar assessed odds. The image of a walk-off home run is vivid, and one that most of us have seen before. Very few of us have read much about dam collapses, let alone experienced one. These are the psychological barriers to taking a probability like 6% seriously.  

Keeping Watch, Taking Action

Right now, Iraqi forces assisted by international advisors are reclaiming territory at the doorstep of West Mosul. If a dam collapse occurred during their push into the center of the city, Iraq’s elite counter-terrorism forces might get wiped out and the Iraqi Army would likely have to abandon all of their equipment. If U.S. and international forces advising Iraqi ones were in the vicinity of the Tigris River during a collapse, those lives might be lost as well. The permanent solution might best be implemented by the Iraqis with the assistance of an international coalition of funders (similar in composition to the countries currently taking the fight to ISIL). President Trump has yet to weigh in on the Mosul Dam issue, but even a 1% chance of a collapse should be taken quite seriously. That GJ Open forecasters think the chances are six times greater should give everyone pause, given the stakes. After all, forecasters can only tell you what they think. It is up to policymakers, both in the United States and in Iraq, to act appropriately.


Welton Chang is a psychology graduate student at the University of Pennsylvania and a Superforecaster for Good Judgment Inc. He is a former U.S. Army officer and Defense Department analyst who served two tours in Iraq. The views presented here are his own and do not necessarily represent those of his current or previous employers. Follow him on Twitter: @weltonchang


Share on FacebookTweet about this on TwitterShare on LinkedInEmail this to someone

7 thoughts on “Forecasting the Future of the Mosul Dam

  1. Depending on who currently holds the dam and its environs, it is not out of the realm of probability that ISIL, convinced it was going to be booted out of Mosul, could make a quick strike, take the dam for long enough to blow a big enough hole that would rapidly be enlarged by water pressure, and accomplish the collapse. ISIL has repeatedly demonstrated that no atrocity is too repulsive for it to perform, so this seems a no-brainer for them if they find themselves losing.

  2. Great analysis. Small nit. Doesn’t a 6% chance convert to around 15:1 odds, not 9:1? (doesn’t change your examples or analysis)

    1. Hi Fang, Thanks for your comment — point well-taken. I’ll contact the author and see if we can get it updated. – Best, Gwyn

  3. I’m a retired civil engineer who specialized in water related projects including flood control. I have no knowledge of the particular factors applicable to the Mosel Dam. But a general observation is that when there are hundreds or thousands of communities throughout the world (or even the U.S ) that have even a small chance of being devastated by a major catastrophe in any year, the chances of such a catastrophe occurring somewhere are quite high. So, in my opinion, the chance of any one occurring and killing thousands of people should be reduced to less than 1 in 1000 per year, perhaps by evacuating people from the hazard zone.

    1. Hi Carl,

      Thanks for your comment. I think that is indeed the hope of improving forecasting systems — that policymakers will take action based on better forecasts. Are you a forecaster? If not, I think you would enjoy it – your analysis quite reminds me of how the Superforecaters think! It take a lot of work and research to succesfully forecast, but thought you might enjoy the challenge. Thanks — Gwyn, Sr. Associate, GJI

  4. I had almost daily interactions with the Ministry of Water Resources for over two years during the American occupation. Ten years ago, late 2007, the US Army Corps of Engineers issued dire warnings to everyone in Iraq that the Mosul Dam was in imminent danger of collapse. They advised people to relocate to higher ground, warned about Baghdad being flooded and thousands being drowned. They ignored the fact that two grouting crews have been working on the dam since its inception in 1981. These dire warnings come from the people that the author has labeled “experts such as the U.S. Army Corps of Engineers”. These are the same experts that built the Teton Dam which was an earthen dam on the Teton River in the state of Idaho in the United States. It suffered a catastrophic failure on June 5, 1976, as it was filling for the first time. The resulting property damage and death claims cost the US Government well over $400 million dollars at that time.

    The peak fill for the Mosul Dam was in 1988 and again in 1991. The Iraqis carefully monitor and control the levels in the reservoir. Contrary to what most Westerners seem to believe, the Iraqis, who have been managing water projects for over two thousand years, know what they are doing. They have no desire to destroy their country and their families. As to ISIL destroying the dam, yes, it is possible to blast a hole in it to rupture the embankment. That would involve a lot of machinery, a lot of labor and a heck of a lot of HE. That process would also be very observable by the satellites and drones that keep an eye on that part of the world. Yes, a flood would result in damage and bodies. The most commonly used bomb is the 500 pounder which does not float. The desert area would rapidly soak up the flood waters and the mud would dry out in a matter of days.

    This is, IMHO, another effort by a bureaucracy to find something to perpetuate their existence. I am confident, based on my personal observations and conversations with Iraqi Ministry of Water Resources personnel, that the Mosul Dam will still be functioning another 10 years from now.

Leave a Reply

Your email address will not be published. Required fields are marked *