Are we dealing with below-normal water levels?

Posted by Sean Fleming

So what exactly is “normal,” anyway? This question isn’t only about trying to decide whether that nose ring your child just came home with is a worrisome behavioural quirk, or something harmless all the kids are doing now. It’s also an important question when it comes to BC Hydro’s reservoir inflows.

While BC Hydro’s profitability depends on many different things, the amount of hydroelectric power that could in principle be generated is obviously tied pretty closely to how much water is delivered by rivers to our power plants, and that varies year to year depending on such things as winter snowpack.

Naturally enough, then, there can be considerable interest in describing inflows accurately, and in lending context to those raw numbers – say, flow volumes in millions of cubic metres – by comparing them with normal conditions. 

But how much higher or lower than normal does an inflow measurement or forecast have to be to really catch our attention as being unusual? How do you define “normal?”  And will there be a new normal with climate change?

A look at Williston Reservoir

Current water supply forecasts call for below-normal inflow volumes at Williston Reservoir this year – but what exactly do “normal” and “below-normal” mean?

A simple and common way of approaching the problem is for the scientific community to agree on a stretch of time and call that the normal period.  The 30 years spanning from 1971 through 2000 are often – though not always – used. The average value of temperature, or precipitation, or river flow is calculated over that period using historical data records. 

It’s easy then to express the current value in terms of that normal value. For example, the water supply forecast issued by BC Hydro in July for February-through-September total inflow volume to Williston Reservoir was 77 per cent of normal – clearly, a low-water year. 

What's normal? What's low?

But how low is low?  We’ve defined normal, but how should we define abnormal? What makes a low or high inflow sufficiently low or high to be noteworthy?

We can approach this “noteworthiness factor” a few ways. 

One is to express the reservoir inflow volume in terms of a ranking. For instance, the aforementioned Peace River inflow volume would (if the forecast pans out) rank as the fourth lowest in the 51 years we've been recording water levels there. 

Another way is to use a statistical measure called the standard deviation. Some rivers fluctuate more than others, just like the prices of some stocks fluctuate more than others. The standard deviation is just a number that helps capture how variable something is, and when applied to the recorded past flows of some particular river, it gives us an idea of how “volatile” (to borrow a word from the finance industry) the water supply in that watershed is. 

So if the current year’s flow is more than one or two standard deviations away from the normal, we have a pretty good indication that we’re seeing something a bit unusual for that particular river. 

Returning again to the Peace River example, the forecast calls for a flow volume that is more than one but less than two standard deviations below the 1971-2000 normal, so we might choose to call it below-normal, but not strongly below-normal.

What tree rings tell us: The Columbia River

The Columbia is one of the world’s great rivers, and the source of much hydroelectric power generation in both Canada and the U.S. Its flow as measured downstream at The Dalles, Oregon is often used as a rough measure of total water supply in the basin.

The graph to the right shows an approximate reconstruction of yearly average flow at The Dalles going back to the mid-18th century.

It was developed by Dr. Ze’ev Gedalof (University of Guelph) and co-workers, from measured changes in tree rings – this is possible because tree growth is related to many of the same environmental factors as river flow. We can see how the hydroclimate of the Columbia Basin has varied over the decades and centuries.

This variation is caused by such things as natural fluctuations in the state of the Pacific Ocean – to which human-caused global climate change may be added now. The chart also shows how defining average or “normal” flows, against which to compare the current year, might be a bit tricky.  Because of the longer-term fluctuations, we would get a different average depending on how much data we use to calculate the average and from when.

Unprecise science

In most practical cases, we have only (at best) a few decades of data against which to compare the current year’s flow, and it may not be clear that those recent decades are the best benchmark. What’s more, many scientists believe that climate has a property called nonstationarity, which would mean that – at least theoretically – a true, fixed benchmark may not even exist.

Still, hydrology and climatology aren’t precise laboratory sciences, and there’s still some room for interpretation here. Because regional climate conditions vary naturally over the decades, if we used slightly different normal periods, we would get slightly different results. The amount of data available in a particular watershed makes a difference too – the fourth-lowest flow in five years, or 50 years, or 500 years would each have very different implications.

Stream flow generally hasn’t been measured all that long in western North America, so it can be tough to tell just how remarkable a record or near-record low or high value really is – or isn’t.  And whether a higher- or lower-than-normal inflow is noteworthy may also depend on context: for instance, more water may be better for one thing but irrelevant (or even problematic) for another. 

In part for these reasons, categorizing inflows – as normal, high, very high, and so forth – is mainly done for communications purposes, with detailed technical work usually focusing on the raw numbers instead.

And what about the long-term global climate changes that are widely believed will result from deforestation and fossil fuel combustion? These may shift the average. What’s “normal” could change over time.

We have to be careful, though, not to ascribe every year-to-year fluctuation to climate change. For instance, preliminary results from research performed for BC Hydro by the Pacific Climate Impacts Consortium suggest that Peace River total annual inflow volume might increase somewhat under future climate changes, so this year’s low inflows seem unlikely to have much to do with the broader phenomenon of global warming.

What’s “normal” when it comes to reservoir inflows? At BC Hydro, we use standard methods to describe what makes the yearly water supply normal, or high, or low for each river. These measures are very useful – but they come with caveats and have to be treated with some care.

Sean Fleming is a senior hydrologic modeller with BC Hydro’s Runoff Forecast Team, and an adjunct professor at the University of British Columbia.