Normal Versus Extreme Hydrologic Events

The immense variability observed in precipitation and streamflow leads one to wonder what actually constitutes an 'extreme' event. For example, most rivers tend to flood (i.e., water completely fills the channel and spills out onto adjacent floodplain) every one to five years. River discharge during such events is often on the order of 10 times the mean annual flow and often 100 to 1000 times greater than the lowest flows. In that context, perhaps they could be considered extreme. However, considering them within the context of all the floods that occur over a century, floods that occur every one to five years would be referred to as 'common floods' (e.g., all the events below ~ 25,000 cfs for the Lehigh River in Figure 4 shown earlier). So, labeling an event as 'extreme' requires some context as to the timescales being considered. Similarly, what is considered 'extreme' varies from place to place. For example, a rainfall event that delivers 5 cm of precipitation is quite rare in Utah, but is nearly a daily occurrence in parts of Hawaii. While there is no formal, universal definition for what hydrologists consider to be 'extreme' events, there are numerous ways we can assess precipitation and streamflow events within the appropriate context (timescale and location) to determine how they compare with 'normal' conditions.

You'll notice that the distribution of flood events in Figure 4 (on a previous page) has a strong right (also called positive) skew, meaning a long tail to the right of the graph. This is commonly observed for flood frequency data. It might be tempting to see the two events that exceed 90,000 cfs and simply label those as extreme events, but for many rivers there is not such a clear cut-off. Instead the most common way that hydrologists determine the rarity of an event is by calculating the frequency with which the event has occurred in the past, and using that as an estimate of the probability that it will occur in the future, as discussed in the example of the Lehigh River above. This is a rational and useful way to make predictions, but note that climate change throws a bit of a wrench into the notion of using the past to predict the future. If the entire distribution of events shifts, so change the probabilities associated with all of the events. We'll address this issue towards the end of the module.

In any case, terms like 'extreme' may be useful for news headlines and catchy titles for scientific presentations, but nature doesn't easily fit into boxes like 'extreme' and 'normal'. Instead, hydrologists tend to use more well-defined terminology to characterize hydrologic events according to their frequency, duration, and magnitude as well as the spatial extent. Events that occur infrequently (i.e., events of low probability) are the ones to watch out for!