Flood Science

COVID-19, Black Swans, and Flood Risk

|

March 24, 2020

Black Swans

COVID-19 has blindsided the world. We are witnessing the unimaginable—tremendous numbers of fatalities in Asia, Europe, and North America, overwhelmed healthcare systems, mass shortages of supplies, and widespread travel restrictions.

Bare grocery store shelves at a Neighborhood Walmart in Biloxi, Mississippi, on Saturday, March 14, 2020. A severe shortage of paper and cleaning supplies was experienced around the country. Photo: Hal Needham

This pandemic and social fallout can be classified as a Black Swan, a rare, low-probability event that was unforeseen and will inflict severe losses on society. The name Black Swan comes from surprised Europeans, who were astonished to find black swans in Australia in the 19th Century. They assumed black swans didn’t exist, as they had only ever seen white swans in Europe. They learned that just because they had not previously seen black swans, didn’t mean they ceased to exist.

A Black Swan is something that blindsides us because we mistakenly assumed the probability of the event was zero. When it occurs, we realize our mistake. Clearly, if something occurs the probability is not zero… it may be a very low number but not zero. Catching us unprepared, Black Swans inflict a disproportionate amount of losses.

Black Swans have caused severe and rapid societal changes throughout history in four major categories: human health, economics, warfare/ terrorism, and natural disasters. Think of how the world suddenly changed from pandemics like the 1918 Spanish Flu, economic collapses like the 1929 Wall Street Crash, surprise attacks on Pearl Harbor in 1941 and New York/Washington in 2001, and natural disasters like the Indian Ocean Tsunami (2004) and Hurricane Katrina (2005).

Burning of the USS Arizona during the attack on Pearl Harbor, December 7, 1941. The Pearl Harbor attack was a Black Swan, an unforeseen event with major global impacts.
Photo:
https://en.wikipedia.org/wiki/File:The_USS_Arizona_(BB-39)_burning_after_the_Japanese_attack_on_Pearl_Harbor_-_NARA_195617_-_Edit.jpg

All of these events happened, in part, because nothing in history matched the nature and magnitude of these events. Before 9/11 terrorists never attempted to hijack domestic U.S. flights and use airplanes as missiles. Before Hurricane Katrina, water levels along the Gulf Coast had never reached such heights in recorded history. Nassim Taleb, the author of The Black Swan, shares how a Black Swan in the form of a brutal civil war shattered the peace of his region of Lebanon after nearly 13 centuries of multi-cultural peace and coexistence. Nothing in the history of this region in Lebanon suggested it was a hotbed for a brutal war (Taleb 2007).

Interpreting Black Swans in severe weather events

Those interested in flood and wind risk should pay close attention to Black Swans because severe weather events clearly fit the description of this phenomenon. Severe storms often ravage communities, where residents cry out, “This has never happened before!”

We can interpret these events in a few different ways.

1) Repeating History

Storms that appear to be novel (new) are often repeating previous weather conditions that occurred in history, but are unknown to the local resident. Keep in mind the one or two highest flood events on record in a location are often considerably higher than all of the other events. Because severe floods are rare events, the highest flood on record in a community may have occurred 100 or more years ago, and most locals are unaware of it.

The graph below shows the flood history of Corpus Christi, Texas, taken from the U-Surge Project (www.u-surge.net). Every flood event on this graph is depicted by a light blue circle, and the slanted blue area near the bottom of the graph shows rising sea levels over time. Note that the highest flood event on record occurred in 1919, and no substantial flood events exceeding 5 feet have occurred since 1980. If a coastal flood matching the 1919 event happened today, most locals would incorrectly feel they are witnessing an unprecedented event.

Time series of coastal flood events in Corpus Christi, Texas. This graph shows the storm tide height for 42 hurricanes and tropical storms since 1900. Each blue circle depicts the maximum water level per event.
 Source: www.u-surge.net

2) Randomness

I walked around the city of Mexico Beach, Florida, the night before it was leveled by category 5 Hurricane Michael in 2018. I remember thinking, “These people have no idea what they’re in for,” and there was truth to that statement. Even if someone was 170 years old, and had a memory of every weather event since official hurricane records started in 1851, there is nothing in the weather history of Mexico Beach that would suggest the possibility of 160 mile-per-hour sustained winds and a 16-foot storm surge on October 10, 2018.

The aftermath of category 5 Hurricane Michael in Mexico Beach, Florida. Sustained winds of 160 mph and a 16-foot storm surge leveled the coast, leaving a scene of devastation.
Photo taken Saturday, October 13, 2018, by Hal Needham.

We need to understand that randomness has a strong influence on the nature of severe weather events. Hurricane Michael could have just as easily struck a different part of the Gulf Coast, and in the next 10,000 years, it’s possible those extreme wind speeds and water levels will never be matched again in Mexico Beach. Yet, hurricane climatology reveals that some areas are more vulnerable than others, so the influence of randomness has its limits.

As I worked on this blog post, I tried an experiment. I cast a six-sided dice 12 times and recorded the results. Each side of the dice had an equal chance of landing side-up. In a perfectly predictable world, each side would have fallen side-up twice in 12 rolls.  Instead, the results looked like this:

Side of the Die Number of Times
1 1
2 2
3 1
4 4
5 1
6 3

Imagine if these six numbers represented six cities along a section of coast with similar flood risk. Perhaps each roll of the dice is a flood event, with one event per year over 12 years. We know each city has the same risk of flooding, but randomness enables City 4 to experience four floods, while cities 1, 3 and 5 each experience one flood. How quickly would the human mind start making “conclusions” about why City 4 is much more flood-prone (the shape of the coastline, atmospheric circulation, too much concrete, etc.) when this city is actually no more vulnerable than the others?

Pensacola, Florida, has never observed category 5 hurricane winds, but the probability of this happening is not zero. The city is situated about midway between Mexico Beach, where category 5 Michael struck, and coastal Mississippi, where category 5 Hurricane Camille struck in 1969. Why have category 5 hurricanes struck cities to the west and east of Pensacola, while sparing the western Florida Panhandle? In part, this is a function of randomness. Given enough time, a category 5 hurricane will strike Pensacola.

Pensacola, Florida. Link here: https://www.flickr.com/photos/timparkinson/4853125181

Would a category 5 hurricane strike at Pensacola be considered a Black Swan? That’s a good question and the answer depends on how we look back at historical data. If we look solely at wind data for Pensacola, we would conclude that category 5 winds were unprecedented and should not be expected. But if we take a more regional look, we see a precedent for such an event, although even within the region, such catastrophic wind speeds would be rare.

3) A Changing World

A third interpretation of Black Swan weather events relates to changes we are seeing in a warming world. We need to tread very carefully on this interpretation, because many people interpret events that fit #1 and #2 above as climate change, when in fact history is repeating itself or we are seeing randomness in the dataset.

We need to identify the change in an underlying physical process to connect an event to climate change. For example, unusually warm water temperatures off the Texas Coast could have contributed to flooding in southeast Texas during Hurricane Harvey (2017) and Tropical Storm Imelda (2019). Onshore winds moving over these exceptionally warm waters likely enhanced the tremendous rainfall amounts in these two storms. This type of connection may point towards the influence of climate change because warmer sea surface temperatures enhance evaporation rates and available moisture in storms. These physical properties point to a mechanism for change beyond randomness.

Flooding during Hurricane Harvey (2017) in Galveston, Texas. Climate change may have enhanced Harvey’s impact, as onshore winds blew over warmer-than-normal Gulf waters for multiple days during this flood event. Photo: Hal Needham

Computer models suggest that climate change could enable “gray swan tropical cyclones” to impact places like Tampa, Florida, Cairns, Australia, and the Persian Gulf in the future (Lin and Emanuel, 2015). The authors use the term “gray swan” as the intermediate ground between white and black swans. Tropical cyclones have either directly impacted these locations before or impacted nearby locations, so future events may not completely blindside local populations. Yet, the nature (intensity) and track of future tropical cyclones may be unprecedented.

By contrast, simply observing multiple severe weather events does not necessarily connect these events to climate change. Look back at the flood history graph for Corpus Christi. Note that the 15-foot storm surge of 1919 occurred just three years after the 9.8-foot storm surge of 1916. These events were clustered close together. If that same cluster of events happened today many people would call it climate change (“I’ve lived here my whole life and never saw this before!”), when, in fact, it appears to be a random clustering of events.

Damage from the 1919 hurricane in Corpus Christi, Texas. A 15-foot storm surge inflicted severe damage on the city, just three years after a 9.8-foot storm surge in the 1916 hurricane.
Photo: https://www.weather.gov/crp/hurricane1919

Moving Ahead

So where do we go from here? How does knowledge about Black Swans help us navigate the world we live in and what are steps we can take to optimize our future while minimizing losses?

Here are three suggested steps.

1) We need to use our imagination.

I love brainstorming, dreaming and imagining. I prefer to focus these efforts on creating or innovating something new.

Unfortunately, in disaster science, we need to imagine every possible thing that could go wrong and every weak link that could unfold in a disaster. While this is not “fun,” it is rewarding to develop a framework that protects our communities from being blindsided.

Fortunately, previous disasters and common sense guide our thinking. When we imagine what could go wrong, we generally want to look at other disasters that have struck our region (country, state, city, county, etc.) and modify them to see what new “angle” a new disaster could take.

For example, we now know that terrorists could use airplanes as missiles, so it makes sense to secure our airports. It pays to imagine what other high-density areas they could strike because terrorists like to blindside society and they are already thinking “outside the box.” Therefore, we should consider protecting shopping malls, sports stadiums, beach resorts and other places with a high population density, in an attempt to protect against a Black Swan. We should also consider potential weaknesses in our system that could be exploited through cyber-terrorism.

We may react to new scenarios by thinking, “Well, that’s never happened before!” However, such thinking sets us up to be blindsided by future Black Swans.  

When considering flood and wind risk, we should identify those locations that are in a “dangerous neighborhood” (i.e. U.S. Gulf Coast) but have not taken a direct hit from a powerful hurricane. This post previously mentioned the potential for enhanced future flood risk at Tampa, Florida, related to climate change, but even without a changing climate, Tampa is likely more vulnerable than local residents realize.

The historical record does not yet show a powerful hurricane striking Tampa, Florida, on a track that would drive a 20-foot storm surge into Tampa Bay, yet computer models suggest it could happen (Weisberg and Zheng, 2006). It just takes some imagination to shift the track of Hurricane Michael (2018) south or Hurricane Wilma (2005) north, to create an unprecedented event at Tampa.

Tampa, Florida. Link here: https://commons.wikimedia.org/wiki/File:Aerial_view_of_South_Tampa,_Florida.jpg

2) We need to interpret events correctly.

If thieves rob your neighbor’s house twice, while leaving your house alone, you can interpret these events in two ways. You could think:

A) “Thank God the robbers only hit my neighbor’s house and left my house alone!”…or

B) “Oh no… I live in a more dangerous neighborhood than I realized! How can I better protect myself?”

While differences in the quality or appearance of these homes (fence, security cameras, etc.) could explain the robbery pattern, the properties of randomness suggest homes of equal value and appearance may observe different break-in histories because of randomness. It is important that we identify randomness for what it is, and avoid forming a false sense of security. Residents who have narrowly escaped storm strikes or floods on multiple occasions should not interpret their good fortune as a long-term pattern that protects them from harm.

Consider the incredible “luck” of Miami-Dade County in South Florida over the past five hurricane seasons. In this time, three hurricanes narrowly missed striking this densely-populated county with hurricane-force winds. While Hurricane Irma (2017) inflicted some flood damage in the county, impacts would have been much worse if the storm tracked farther east, and none of these storms produced hurricane-force winds in the county.

Hurricane-force wind swaths near South Florida for the years 2016–2019. These swaths depict the approximate area of hurricane-force winds from Matthew (2016),  Irma (2017), and Dorian (2019).
Source: Brian McNoldy, University of Miami. Link: http://bmcnoldy.rsmas.miami.edu/

Such climatology should be sobering for residents of South Florida, but too often people misinterpret such randomness and good luck as a long-term pattern of protection. (I lived in Miami-Dade County for six months in 2018, and formed this opinion after dozens of conversations with locals.)

3) We need to be prepared.

Finally, be prepared for the unexpected. Black Swans inflict high losses because they blindside us and we cannot predict them far in advance. We will never be able to perfectly predict Black Swans.

As the year 2020 arrived, who would have predicted that hand sanitizer and toilet paper would be impossible to find? Who could have imagined that most schools around the country would close their doors?

The future is hard to predict, but there are some basic ways we can protect ourselves.

A) Set aside emergency supplies. This includes food, water, cash, and basic medical supplies.

A basic supply kit can be surprisingly cheap, but many people do not have one. A few years ago, I realized that I travel the country giving talks on disaster preparedness and I did not have adequate food and water reserves for three days. I changed this and created a kit with emergency supplies to last me one week for less than $20. If you think about it, a gallon of water in most parts of the country costs less than one dollar and canned food is relatively cheap.

B) Create multiple safety nets.

Think about the technologies you use to communicate with friends and family or store your most important documents and memories. Imagine if these technologies were gone. This doesn’t have to be in some coordinated cyber-terrorist plot, you could simply drop your smartphone in water and then it is lost to you.

Consider backing up your data on the cloud and writing down important phone numbers on paper. Back up important files in multiple sources, and make as many safety nets as possible for your life. Even if your house has never flooded before, consider storing important documents and family treasures at higher levels of your home in case a flood ever strikes.

C) Give yourself a buffer.

Finally, give yourself a buffer when preparing for natural disasters. Expect a future flood could inundate your community with at least several feet more water than the highest flood on record. Consider elevating your home at least several feet above the Base Flood Elevation (BFE), or purchasing a house farther from the flood zone if given a choice. Consider adding additional insurance coverage if your plan provides that option.

We can reduce our losses substantially when we build in a buffer of security instead of walking to the edge of the proverbial cliff. By following these basic principles, we can protect ourselves against a Black Swan, and reduce potential losses from unexpected future events.

References

Lin, N., and K. Emanuel, 2016: Grey Swan Tropical Cyclones. Nature Climate Change, 6, 106-111. Link: https://www.nature.com/articles/nclimate2777

Taleb, N.N., 2007: The Black Swan. Random House Publishing Group, New York, New York, USA. 366pp.

Weisberg, R. H., and L. Zheng (2006), Hurricane Storm Surge Simulations for Tampa Bay, Estuaries and Coasts, 29, 899 – 913.