top of page
All hurricanes create a crisis. Not every crisis is a hurricane.
All hurricanes create a crisis. Not every crisis is a hurricane.

Technical professionals rightly take pride in accuracy. Accuracy is the foundation of good engineering, sound science, and a responsible operation. However, in a crisis, accuracy alone does not meet the needs of senior management, regulators, or the public. People want to understand what is happening. They want to know how the events impact them, their families, and their survival. A technically accurate message that does not address these concerns will fall short.

 

Crisis Communication

Crisis communication is not a technical exercise. It is a leadership function. It requires context, empathy, and relevance. After all, no one wants to be lied to in a crisis. But people process information differently when they are under stress. They care most about their own survival and absolutely nothing about how right you were (or how good you think you are).

 

“No one wants to be lied to in a crisis.”

 

Technical accuracy is necessary, but it is not sufficient for effective communication.

 

From the Real World

That Great Flood of 2017 hit New Orleans unexpectedly on August 5th. The next week more flash flooding occurred. The Executive Director of the Sewerage & Water Board blamed it on climate change. Others said it was because only 38 of 58 of the below-sea-level city’s drainage pumps were operable.

 

I found myself dispatched to New Orleans in mid-August to evaluate the system and help get it back up to speed. Technically, my role was reliability, while two other colleagues led the physical improvements for the pumps and the electrical system.

 

Less than a month later, what would become Hurricane Harvey came barreling across the Gulf, looking for a landfall somewhere between New Orleans and Brownsville, Texas. We were not ready.

 

In the week preceding the storm, my job was heads-down, focusing on what was broken and assessing (day-by-day) the realistic performance of the drainage system. One part of the August debacle was that no one accurately knew what the drainage system could do. The other part was poor crisis communication.

 

By Wednesday morning of that week, I was shifted to the emergency management system. The focus had become supporting crisis communication rather than the normal reliability improvement role. My job was to make sure what we said was accurate; however, the communication and the numbers we gave to the public would have a different purpose.

 

By Thursday night, Hurricane Harvey’s path was more definitively leaning toward a Texas landfall. By Friday morning, New Orleans had dropped from the probability cone.

 

“You better get to the airport,” our program manager said.

 

“Yeah, luckily we will not be in crisis mode over the weekend,” I replied.

 

“I think everyone needs a couple of days’ rest. Shifting gears and preparing for the storm was a tough exercise.”

 

“I’ll be back Monday by 9,” I stated as I headed for the door. “The job doesn’t change, but the delivery certainly was going to be different if we got hit.”

 

Accuracy Without Context Creates Confusion

Technical professionals often respond to crises by providing detailed explanations. We describe the data, the measurements, and the system limits. Most of us assume that more information will create more understanding. In reality, more information often creates more confusion.

 

People in a crisis are not looking for technical detail. They are looking for meaning. They want to know whether they are safe. They want to know whether the situation is under control. They want to know what the organization is doing. When the message focuses on technical accuracy without providing context, the audience is left to interpret the information on their own. This leads to misinterpretation and unnecessary concern.

 

Accuracy Without Empathy Creates Distance

Technical professionals sometimes avoid acknowledging emotions because we fear it will undermine credibility. Maybe worse, as analyticals, many of us don’t have a lot of natural empathy.

 

Facts without empathy create distance. People interpret it as indifference or defensiveness.

 

Empathy requires recognition, not emotional language. Mentally putting yourself in the place of one who has been impacted makes the technical message more effective.


Facts without empathy create distance.

 

Accuracy Without Action Creates Frustration

In a crisis, people want to know what the organization is doing. They want to know what steps are being taken to evaluate, stabilize, and resolve the issue. Returning to the normal state of affairs is the normal disaster recovery expectation.

 

A technically accurate message that does not describe the actions being taken leaves organizational and community leaders without the information they need to move forward. That’s a big reason technical professionals need to be present and inform the crisis communication team, but we techies should expect to take a back seat to the crisis communication team. At least until things stabilize and get back to normal (whatever and however good or bad that normal is).

 

Accuracy Without Boundaries Creates Risk

Technical professionals sometimes provide more detail than the situation requires. We share a lot of stuff in case someone needs it. Plus, at least for engineers, our duty-based ethics training tells us to tell our decision makers everything we know.

 

The real problem in a crisis is that what we know is changing. What’s an accurate statement this morning is inaccurate in the afternoon. In other words, early information often changes. When it does, the organization appears inconsistent.

 

Crisis communication requires boundaries. It requires discipline about what is known, what is unknown, and what is still being evaluated. It requires clear statements about the status of the information. This protects the organization and maintains credibility.

 

Boundaries do not limit transparency. They strengthen it by ensuring that the information shared is reliable.

 

From the Real World: Hurricane Hugo

It was early in my career, but I knew this was not a pretty after-action debrief. Hurricane Hugo had just flattened Charleston. The monster hurricane had grazed Myrtle Beach, albeit on the more impacted northern side. Beachfront condos and houses were temporarily condemned. Our job was now to inspect the damage and declare which units would be classified as habitable or uninhabitable.

 

“Did I say something wrong? Did I say something inaccurate?” asked one of my engineering colleagues after speaking to a local citizens group.

 

“Not necessarily inaccurate,” shrugged our boss. “But certainly insensitive.”

 

“Well, the truth is the last 24-hour forecast was correct. And it could have been much worse,” countered my determined engineering colleague. “They have to accept that, and they have to accept that things are not going to get better for months, not weeks.”

 

I remembered the team lecture that followed throughout my career. Our boss made quite an impression on me about what victims “have” to accept, how much impacted people really care about how accurate our forecasts are, and our personal opinions about how fast (or slow) things will improve.

 

In a crisis, context and empathy matter a lot more than defending technical accuracy.

 

Five Steps to Prepare for Crisis Communication This Hurricane Season

Technical professionals do not have to wait for the next storm to put these principles into practice. Before hurricane season is in full swing, take these five steps:

 

  1. Verify the accuracy of your system maps and online equipment information.


  2. Verify your role with the crisis communication team.


  3. With hurricane season here, review local guidance — most counties and states in hurricane-prone regions publish it.


  4. Review CDC and FEMA protocols.


  5. Understand the difference between a crisis communication approach and an approach such as FINESSE. (FINESSE is not a crisis communication tool.)

 

Accuracy Must Be Part of a Larger Structure

Technical accuracy is essential, but it must be integrated into a communication structure that supports decision making. That structure includes context, empathy, action, and boundaries. It helps leaders understand the situation. It helps the public feel informed. It helps regulators see that the organization is acting responsibly.

 

In a crisis, accuracy is an important foundation. However, understanding the communication context and empathy will carry the day. You create effective communication that supports leadership and strengthens trust when you combine accuracy and empathy.

 


This article first appeared on Substack: Solomon, J. D. (2026, August 12). Why technical accuracy is not enough in a crisis. Communicating with FINESSE (Substack). https://communicatingwithfinesse.substack.com/p/why-technical-accuracy-is-not-enough


References:

WWLTV. (2017, November 15). Timeline: How New Orleans flooded on Aug. 5, 2017 [Video]. YouTube. https://www.youtube.com/watch?v=FKTuUzTUgXA

 

Solomon, J. D. (2020, August 18). Overcoming crisis caused by rare events. https://www.jdsolomonsolutions.com/post/overcoming-crisis-caused-by-rare-events

 

Solomon, J. D. (2021, November 8). Success or failure determined by addressing rare events. https://www.jdsolomonsolutions.com/post/success-or-failure-determined-by-addressing-rare-events

 

Solomon, J. D. (2022, September 27). Improve your communication of rare events before they happen. https://www.jdsolomonsolutions.com/post/improve-your-communication-of-rare-events-before-they-happen

 

Solomon, J. D. (2023, February 11). Improve your career by showing empathy to the decision maker. https://www.jdsolomonsolutions.com/post/improve-your-career-by-showing-empathy-to-the-decision-maker

 

Solomon, J. D. (2023, March 6). One question creates empathy and improves business presentations. https://www.jdsolomonsolutions.com/post/one-question-creates-empathy-and-improves-business-presentations

 

Solomon, J. D. (2024, January 29). How context matters in successful asset management implementation. https://www.jdsolomonsolutions.com/post/how-context-matters-in-successful-asset-management-implementation

 

 

JD Solomon writes and consults on decision-making, reliability, risk, and communication for leaders and technical professionals. His work connects technical disciplines with human understanding to help people make better decisions and build stronger systems. Learn more at www.jdsolomonsolutions.com and www.communicatingwithfinesse.com.

2025 was the first hurricane season in which the National Hurricane Center formally tested AI guidance alongside its traditional tools.
2025 was the first hurricane season in which the National Hurricane Center formally tested AI guidance alongside its traditional tools.

AI is not the story of the 2026 hurricane season, and it was not the story in 2025 either. However, the technology is real and the results are improving fast. Plus, every major forecasting center is now testing AI. My experience with AI and hurricane forecasting tells me AI will become a primary tool in hurricane forecasting next year or in 2028, but not now. Here’s why.

 

The AI Trend Is Real

Like other applications of AI, weather models have moved from research curiosity to actively using AI in just a few years. Google DeepMind's Weather Lab, which I added to my Top 6 list this year, is the most visible example. DeepMind's AI model runs at roughly 0.3% of the computational cost of a traditional forecast and can generate a thousand possible storm paths instead of the usual fifty.

 

NOAA has been building and testing AI versions of its own forecast models alongside the models themselves. 2025 was the first hurricane season the National Hurricane Center formally tested AI guidance alongside its traditional tools. The investment has been aggressive since.

 

That AI trend is up and to the right; however, that is not the same as an established track record. Hurricane forecasting is not the place to confuse the two.

 

How Forecasters Get It Right Today

Today, most of the hurricane forecasting heavy lifting comes from physics-based models. Real observations, gathered by hurricane hunter aircraft and satellites, feed a supercomputer that simulates how the atmosphere will actually move and change.

 

A second, smaller set of models leans on history rather than physics, essentially asking how storms with similar traits have behaved in the past. NHC then blends all of that guidance. A meteorologist applies judgment on top.

 

NOAA NHC hurricane forecasts do a good job of predicting landfall locations and windspeed within 48 hours of landfall.

 

NHC's official forecast beats every individual model that feeds it, because a person is doing the blending. Built on decades of refinement, that human-plus-physics combination is the world my Top 6 hurricane sites are built around.

 

What AI Actually Adds

The natural question is whether AI is helping because it works with better data or because it computes faster. Right now, it is overwhelmingly about speed.

 

AI models do not need more detailed data than physics-based models. Based on what I have seen, they hold up fine on coarser data. What they need far less of is computing power.

 

The forecast that once took hours on a supercomputer now runs in minutes. That speed is what allows a thousand-storm-path ensemble instead of fifty.  A bigger ensemble is a better picture of what's uncertain, not proof the uncertainty is gone.

 

AI is not out-sensing or out-observing the traditional models; it is making the existing forecasting pipeline dramatically cheaper to run at scale.

 

Accuracy is uneven and depends heavily on the history a given model was trained on. Some AI models track storms well but forecast intensity poorly, or the reverse. They recognize patterns from the past, so they struggle with storms unlike anything in their training data. A University of Chicago study made the point directly: researchers trained a model on data with no major hurricanes, then asked it to forecast one. It failed.

 

That is not only a data problem. People have the same blind spot. We tend to overweight the rare events we can picture and underestimate the ones we can't, which is exactly why the storms nobody saw coming are the ones that do the most damage – to a training set or to a coastline.

 

What It Means for You This Season

Within the seven-day window that determines whether you batten down the hatches or evacuate, stick with NHC's official forecast and the consensus-based sources on my Top 6 list, not any single model, AI included.


It also helps to know what the AI-versus-traditional race is actually about. Both sides are chasing better track and intensity forecasts, and within about 48 hours of landfall, the existing models already do that job well.

 

 The forecast problem that still kills the most people is different: rainfall and inland flooding, which neither the traditional models nor the AI challengers have solved. A storm can be forecast almost perfectly and still be devastating for reasons the cone never showed. A hurricane is rarely one event anyway – it's wind, then surge, then the inland flooding that does its damage days later and hundreds of miles from the coast.

 

For the Real World (As A Practitioner)

In my consulting work, clients responsible for critical infrastructure increasingly ask whether an AI forecast is good enough to plan around. I think my answer is the one a meteorologist would give: not yet, and not alone.


Right now, AI's value is mostly behind the scenes. It gives NHC forecasters another independent viewpoint and generates larger, cheaper ensembles for research. It has not yet earned a seat at the table for frontline operations decisions that actually matter, like when to shut down your facility, when to secure safe temporary housing for your essential O&M staff, or when to order a tanker of backup fuel for your emergency generators.

 

The Future of AI Hurricane Forecasting

AI's role in hurricane forecasting will continue to grow. But 2026 is another year of testing and refinement, not a changing of the guard. Judgment compounds as models grow, just as it does for human forecasters.

 

My money is on 2027 or 2028 before AI carries significant weight as a stand-alone. But even then, the primary order of business will not be trusting an AI-based tool but rather when to use it or not.  Until AI judgment is built and proven, when a storm is within a week of landfall, trust the forecast that already has a person's judgment behind it.

 


JD Solomon writes and consults on decision-making, reliability, risk, and communication for leaders and technical professionals. His work connects technical disciplines with human understanding to help people make better decisions and build stronger systems. Learn more at www.jdsolomonsolutions.com and www.communicatingwithfinesse.com.

A new category for AI sites, but traditional sites are still the go-to in 2026.
A new category for AI sites, but traditional sites are still the go-to in 2026.

Hurricane season is upon us in the US Southeast. Once again, you’ll need the best sources of information to tell you when it’s time to batten down the hatches or get ready to be on the move. This year, I’ve added a new category on AI forecasting models, along with updated notes on emergency flood models, seasonal forecasting, and ocean temperature trends. These are my Top 6 sources for predicting and tracking hurricanes in 2026.


#1 NOAA National Hurricane Center (NHC)

Still the most authoritative source for real-time data, forecasts, and advisories. New for 2026: smaller, more accurate forecast cones and an experimental cone that shows the true range of track uncertainty. A must-have on any list.


#2 Tropical Tidbits

Levi Cowan continues offering excellent insights, making this site a favorite among weather enthusiasts and professionals.


#3 Mike’s Weather Page (Spaghetti Models)

This page remains a valuable resource, especially with its aggregation of multiple models and easy-to-digest updates.


#4 Track the Tropics

The site is a comprehensive resource with quick access to various models and data.


#5 Weather Underground

Although it has undergone some changes over the years, it remains a strong source of localized weather data.


#6 The Eyewall

The Eyewall – Seeing you through the storm



Emergency Response

North Carolina has a great publicly available site called FIMAN (Flood Inundation Mapping and Alert Network) for travel needs. FIMAN displays current and forecasted water levels from hundreds of locations throughout North Carolina’s streams and coasts, helping residents and travelers make informed decisions during flood events.


Some Southeast U.S. states, including state departments of transportation (DOTs) and emergency management systems, have developed similar tools in recent years. If you are traveling after a major weather event, find your local sites.


Seasonal Forecasting

WeatherTiger

WeatherTiger is boldly reimagining seasonal forecasting and agricultural meteorology through the use of proprietary technologies. It's on Substack and subscription-based. I use it more as a reference and for weather insights than for real-time tracking.


Tropical Storm Risk (TSR)

Tropical Storm Risk (TSR) offers a leading resource for predicting and mapping tropical storm activity worldwide. TSR has won two major insurance industry awards: the British Insurance Awards for Risk Management (2006) and London Market Innovation (2004). Useful early in the season or before storms form; less useful when a storm is active and you need immediate decisions.


AI Forecasting

Google DeepMind – Weather Lab

Weather Lab is Google DeepMind’s free, public site for exploring AI-generated cyclone forecasts from its WeatherNext models alongside the official NHC track. It's new to my list this year and it's become a legitimate complement to the traditional model guidance above – worth bookmarking, not a replacement for the NHC.


Ocean Temperature Trends

University of Miami – Rosenstiel School of Marine, Atmospheric, and Earth Science

The overarching focus of the Rosenstiel School’s Upper Ocean Dynamics Laboratory (UODL) is to develop, implement, and disseminate academic research on the coupling between the oceanic and atmospheric boundary layers under both weak and strong wind conditions.



University of Maine – Climate Change Institute

Climate Reanalyzer began in early 2012 as a platform for visualizing climate and weather forecast models. Site content is organized into three general categories: Weather Forecasts, Climate Charts, and Research Tools.



JD Solomon resides in the Carolinas, where he fishes, sails, and coaches baseball. Professionally, JD Solomon is the founder of JD Solomon, Inc., the creator of the FINESSE fishbone diagram®, and the co-creator of the SOAP criticality method©. JD has weathered many storms, both on land and at sea.


Experts
bottom of page