The history of weather forecasting dates back to the history of humanity itself. For millennia, people have tried to forecast the weather. Jesus himself says in the New Testament:

“When evening comes, you say, ‘It will be fair weather, for the sky is red’, and in the morning, ‘Today it will be stormy, for the sky is red and overcast.’ You know how to interpret the appearance of the sky, but you cannot interpret the signs of the times.”

Considerable improvements have been made thanks to modern technology, in weather forecasting. However, hurricane dynamics are considerably more complex compared to regular weather prediction. Hurricanes occur above warm large water bodies. When the temperature reaches above 80 F or so, warm air rises up.

This in turn causes a low-pressure region at the ocean surface, which sucks in surrounding cooler air. The rotation of the Earth leads to a Coriolis effect, causing the warm air moving upward to start spiraling into a hurricane. This hurricane becomes a giant high-speed circulation around a low-pressure ‘eye’. Greedy for moist ocean air, the hurricane sucks up the air in its path and causing havoc along the way, many times due to the dangerously large wind speeds.

U.S. governmental agencies like the National Oceanic and Atmospheric Administration(NOAA), integrate multiple data sources to predict the occurrence and dynamics of hurricanes. These data include precise air temperature and pressure measurements from satellites, balloons, ships, sensors at specific locations.

These data are input into models that use supercomputers to solve mathematical equations governing the physics and motion of the atmosphere in combination with previous data from historical hurricanes.

Hurricane predictions are shown by ‘cone’ types of maps that indicate the most probable area where forecasters expect the storm to arrive. In the last decade, there have been significant improvements in hurricane forecast accuracy.

Apart from the hurricane itself, there are a number of additional impacts that impair the safety of people trying to flee the hurricane, or hunker down and wait for the storm to pass. These include dangerous traffic jams during mass evacuations, sudden large-scale power loss, gas shortages, etc.

While there have been improvements in predicting the path of hurricanes, this does not necessarily translate to mitigating the impacts on people and helping those in distress safely navigate dangerous conditions.

Lives are at risk due to life-threatening conditions during disasters, that sometimes quickly cause complex societal disruptions that we might never have expected previously.

This is where big data and analysis techniques have a huge untapped potential to build community resilience to hurricanes.

Streamlining Evacuations

Just 3 weeks after Hurricane Katrina, Hurricane Rita hit coastal Texas. Officials strongly encouraged residents to evacuate. What ensued was the largest evacuation in the history of the United States — with more than 3 million people fleeing prior to Rita’s landfall. During Hurricane Rita, more people died during evacuating due to the combination of excessive heat and gridlock, than due to the Hurricane itself.

Sprawling traffic jams during Hurricane Rita evacuation | Brett Coomer/Houston Chronicle

More than a decade later, officials did not issue evacuation orders during Hurricane Harvey — which also impacted coastal Texas. This decision came under scrutiny as thousands of people that did not evacuate — needed to be rescued from floodwaters.

These 2 diametrically opposite evacuation scenarios in the same geographical location show that neither response was ideal. Pacing traffic and prioritizing evacuations could help reduce risks of mass evacuations, while at the same time getting those at most risk — away from imminent harm.

Google Maps recently added a disaster navigation tool to help route users away from areas with extreme events. However, there could be more potential to efficiently routing users. Research using GPS location data of travelers has shown that coordinated routing during rush hours, could reduce overall travel times.

Accurate knowledge of individual trips can also be used to simulate complex urban traffic patterns, as I’ve shown in this article.

Traffic simulations using SUMO | Skanda Vivek

While Google Maps offers users a way to navigate around disasters, it does not harness the power of coordinated response. One can imagine an app that tells each user a different route (even if they start from essentially the same location), such that the total travel time for all commuters is reduced. The app might even prioritize evacuations, and ask those users in regions that are not in immediate harm to evacuate at another time.

Power and Emergency Outages

During Hurricane Ida, one of the major consequences was the loss of power in New Orleans due to wind damage. In an interview with NPR, the mayor of New Orleans stated that this was due to the damage to power transmission lines outside of New Orleans.

These transmission lines, however, fed distribution lines inside New Orleans, and their outage resulted in a large-scale loss of power in New Orleans, that lasted many days after the original hurricane. An article by the New York Times states that the suffocating heat, caused by this power shortage in fact was the greatest killer in New Orleans and not the Hurricane itself.

During desperate times, it is extremely important that emergency response services like 911 provide timely help to citizens in need. However, during Hurricane Ida, the New Orleans 911 center had a service outage from the power outage. Ultimately, the center was down for 13 hours.

Entergy, the company that provided much of the power for New Orleans had recently opened a new natural gas plant, saying it would help keep New Orleans resilient during intense storms. However, the power was lost because of the failure of components outside New Orleans.

This shows how important it is to consider interconnected infrastructures and have a holistic picture of cascading vulnerabilities during disasters. Data from disasters like Hurricane Ida could help develop resilience in the future. The company poweroutage.us provides information on how many customers are without power based on county.

While this information is good for news outlets documenting the large-scale power loss during disasters, it doesn’t convey the rich interconnected nature of the power grid, and how failures in certain parts due to disasters can lead to broader power outages outside the initial disruption location.

This is where detailed data from various energy providers on the infrastructures that went down, and how this cascaded to other parts of the power grid could be useful in understanding how to build power grid infrastructures that are more resilient to natural disasters.

Drawn Together

There are multiple ways that big data can significantly improve disaster response. We do incredibly well to predict events in isolated systems — such as what path would a hurricane take? Or which areas are at more risk for flooding?

However, we perform incredibly poorly as a society to predict or simply be prepared — for consequences outside the original system. Questions such as how might a hurricane lead to power outages? Or how would extreme cold events cause failures in energy infrastructures (as seen during the recent 2021 Texas Winter Storms) seem to fail us.

Multiple disasters have shown that many times it is not the disaster itself that is the most dangerous. During the case of Hurricane Ida, more lives were lost due to the suffocating heat and power outages. In the case of Hurricane Rita, more lives were lost due to the sweltering heat while stuck in incredibly long traffic jams.

This failing as a society is I believe due to the fundamental nature of the way we reward contributions and penalize failings. The NOAA is tasked with accurately forecasting the hurricane “cone,” and not the power outages resulting from hurricanes. A better hurricane prediction is rewarded.

Whereas New Orleans officials are not penalized for the failings of the power grid outside the parish, even though this resulted in power losses within New Orleans. If no one takes the blame, it becomes no one’s responsibility to ensure that the cascading impacts of natural disasters are limited.

Ultimately the ones that are harmed are people. But paradoxically, we don’t seem to reward holistic, coordinated solutions that minimize risk and maximize the social good. This is where I think big data could play an important role. Hopefully, in the next decade, we see a few of these solutions being realized to build resilience against disasters through a more streamlined response.