Explainer: Is climate change the cause of the 2019 Ottawa River flooding?



Lots of people struggle with the idea that our current flooding may be caused by climate change. For example Ottawa city councillor Eli El-Chantiry on CBC with Robyn Bresnahan said:

“Quite honestly — I’m a little bit — start to think about — like they told us it happened once in a hundred years. Now it happened in two years. So it’s gotta be something has changed. And what caused that change and what we need to change as well. And I know somebody says it’s climate change. I get that, it’s climate change. It has not much changed the climate on us in the last two years. So the first time you had that type of flooding was in 90 years; exactly 89 years and now you’re having it again in two years. So, has something severely changed? I think — I don’t know, I don’t have the answer. All what I can tell you, all communities from here to Temiskaming, they’re at risk, and all the way to Montreal, so it’s not just our area specifically. There’s something really — some, and some answer — had — I don’t have the answer for that question.”

Councillor El-Chantiry isn’t alone in wrestling with this one.

When we hear that a weather event is a once in a 100-year event we shouldn’t think “ah, we had a flood last year, now we’re safe for 99 years” any more than we would think “heads or tails, there’s a 50/50 chance, I got heads last time, so I’m certain to get tails this time.” That 100-year weather event is an expression of probability. There’s a one percent chance of such weather in any given year.

Probability is confusing. Even over 100 years there isn’t 100% chance of that 100-year weather event. According to the US National Weather Service Flood Return Period Calculator in a 100 years there is only a 63.4% chance of a 100-year weather event.

How those probabilities are calculated are from records of the past. But records don’t exist for every aspect of weather for every place or for all that many decades into the past. So probabilities are calculated from the partial records that do exist. The US National Oceanic and Atmospheric Administration site Weather.gov has an explainer on how partial data is used to produce probabilities of 100-year or even 1,000-year events.

Borrowed from NOAA explainer

In short, the partial data is plotted on a graph and a curve is generated that fits the data. New data might change the shape of the curve. There is always some sort of weather so the area under the curve adds up to 100% probability. It’s a little bit of the unlikely plus a larger dollop of the more likely. The fat part in the middle of the curve is the most likely and the skinny part off to the right, where the curve is getting flatter and flatter, is where things are less and less likely. The 100-year event corresponds to the place where there’s only 1% of the area under the curve out there to the right.

All this is true without climate change.

The prominent climate scientist James Hansen has compared climate change to loading the dice of extreme weather probability so that extremes are becoming more likely.

Spring temp increases in Ontario & Quebec

When we hear weather forecasters talk about a given day’s weather and compare it to “normal” they are using a figure for “normal” that is an average of 30 years for  that day’s date. Climatologists have long known that “normal” changes over time and so every ten years they update “normal” to be based on the most recent 30 years.

The 2019 Canada’s Changing Climate Report tells us that Ontario and Quebec have both experienced warmer spring temperatures and increased winter and spring precipitation since 1948.

So for the Ottawa River watershed there is on average more snow in the winter, melting more quickly in the spring, accompanied by more spring rain; three important factors that influence flooding.

Some of these data points will have been accounted for in the 100-year probability forecasts, but not all; and it is the more recent data that are least likely to have been included while at the same time being the most likely to represent a change from the historic values. That doesn’t mean floods every year, but it does mean a new shape to those curves fitted to the data. It does mean water levels that once had 1% probability now have something higher.

For decades scientists have hated the question “was this event caused by climate change.” The careful answer has been that climate change increased the likelihood of it happening. But in recent years some scientists have begun calculating how much the likelihood has increased due to climate change. This is called “attribution” and instead of avoiding saying an event was caused by climate change, they can say climate change made the event a certain percent more likely to happen.

The Canada’s Changing Climate Report includes projections. Under certain scenarios* extreme precipitation events expected once in 50 years may become once in 10 year events.

So if we’re seeing an increased frequency of flooding of the Ottawa River, we’re seeing the beginnings of what climate science predicts. It’s a pretty subtle difference but the reason we can’t say climate change caused this flood is that the changing climate is analogous to loading dice to come up six more often, but that doesn’t imply that unloaded dice never come up six.

6 Comments on “Explainer: Is climate change the cause of the 2019 Ottawa River flooding?

  1. *What I mean by “certain scenarios” is what’s calculated to be a “worst case” scenario. Also called “high emissions” or “business as usual.” But that worst case simply represents a continuation of the greenhouse gas emission pattern that we’re following now, and have been for the 40 years or more that we’ve known it to be a problem. The 50 to 10 year probability change is calculated to accrue over this century.

  2. I am not disputing your scenarios or findings at all, I do believe them in regards to weather changes and temperature forecasts . What your scenarios do not take into account is the act that the Ottawa River is a completely managed waterway from beginning to end. All of the reservoirs and almost all the water passes through multiple dams and Hydro stations all controlled by Hydro One and Hydro Quebec. Since Hydro One went public in 2015 the Ottawa River has flooded twice. This should be looked into a little more closely.

    • My view on this is that complex events are never things that can all be blamed on one thing. I assume managing a watershed is complex thing to do and the changing variables that come with climate change make that even more challenging. A view from a non-expert in these things.

  3. I think I see your analogy here. Thanks for the nuanced explanation. Something I’m wondering though. You have an example where a 100-year event has a 63.4% chance of occurring over 100 years. Does that same principle apply to 50 and 10 year events over similar time periods? So, for example, would a 50-year event have a 63.4% chance of occurring over 50 years? I would assume that it probably isn’t exactly 1:1 but would it be roughly the same, higher, lower?

    Where I’m going with this is, if under certain scenarios a 50-year event moves to a 10-year event it would mean that that 10-year event may not occur, but is still more likely to occur. So climate change may not cause an event to occur but it is increasing the likelihood/risk of it occurring. Am I correct?

    • I think what you’ve said is generally correct. Think of flipping a coin. 50 50 chance of getting heads. But two tosses isn’t 100% likelihood of getting heads. Using that flood return calculator here https://www.weather.gov/epz/wxcalc_floodperiod I get close but not identical results. A 50 year flood has a 63.6% chance of happening over 50 years; a 10 year flood has a 65.1% chance over 10 years.

  4. It’s unfortunate that someone decided to call the event with a 1% annual probability of being equalled or exceeded the event with a return period of 100 years (Return Period = 1/0.01), and then it was mislabelled as the 100-year event. Charles, your article is very good. It should be noted that the probability distribution, the curve, is based on a limited number of measured events, so as more events are recorded, the curve changes. For this reason, we calculate the confidence limits, which are a measure of uncertainty in the estimates.

<span>%d</span> bloggers like this: