At 6:00am UTC on 22 October 2015, Hurricane Patricia was tracking along Mexico’s western coast as a category 1 storm with 39 m/s (138 kph/86 mph) winds. Twenty-four hours later, Patricia’s maximum wind speed had increased to 93 m/s (335 kph/207 mph) and the surface pressure inside the eye had plunged to 886 millibars. That unprecedented intensification prompted CNN to refer to Patricia as “the strongest hurricane … the National Hurricane Center (NHC) has ever had to forecast.”
Indeed, Patricia (seen at its peak in figure 1) became the strongest hurricane in recorded history.
Hurricane Patricia is fascinating not only because of its sheer intensity but also because of the failure of any operational weather model to predict how rapidly the storm would intensify. Figure 2 demonstrates the gross underestimate of Patricia’s intensity by a suite of models. Despite the real storm’s maximum wind speed ultimately climbing to 95 m/s, neither the models nor the official NHC forecast (purple) had predicted wind speeds above 52 m/s.
The models’ significant underestimation of Patricia highlights that although forecasters are getting better at predicting storm tracks, they still struggle with projecting hurricane intensification. That’s a big problem, considering that issuing early, trustworthy warnings to protect lives and property depends on accurate intensity predictions. Recently, at the National Center for Atmospheric Research in Boulder, Colorado, my colleague, Falko Judt, and I demonstrated that it is possible to predict episodes of rapid intensification like the one that turned a relatively tame category 1 hurricane into the strongest tropical cyclone on record. The key is using a high-resolution model with the ability to both accurately represent the structure of the hurricane during model initialization and reproduce environmental factors that correlate with short-term strengthening. We hope our findings will lead to improvements in the intensity forecasting of future events.
Patricia’s evolution began in south Mexico as a series of wind events in mid-October 2015, though the storm was not classified as a tropical disturbance until 20 October. Patricia officially attained hurricane status on 22 October and then intensified dramatically. Based on estimates from Hurricane Hunter aircraft observations and satellite data, Patricia attained a maximum wind speed of 95 m/s (342 kph/212 mph) and minimum pressure of 872 mb at 12:00pm UTC on 23 October. That day dozens of Mexican coastal towns, including Puerto Vallarta and Manzanillo, were declared states of emergency. Officials passed out sandbags, schools were closed, and the recently upgraded category 5 storm was called an “extremely dangerous, potentially catastrophic hurricane.” All the while, models were underestimating Patricia’s intensity.
Fortunately for those coastal communities, Patricia weakened not long before landfall, in part due to vertical wind shear: Winds at different altitudes in the atmosphere started to blow at different strengths, which disrupted the storm’s ability to convert heat into the kinetic energy of its swirling winds. Reaching land as a category 4 hurricane, Patricia endangered people and property more because of the flooding it caused than because of high winds. By 3:00am on 24 October, Patricia was already below hurricane strength. Yet the storm still caused damage totaling $325 million, led to six deaths, and impacted more than 10 000 homes and 100 000 acres of farmland.
In retrospect, Patricia’s rapid intensification proved to be the biggest challenge to prediction models. Rapid intensification for tropical cyclones in the eastern Pacific refers to a period when a storm’s maximum wind speed increases more than 18 m/s within a 24-hour period. Patricia’s increased about 55 m/s. Traditionally, rapid intensification studies have centered on the Atlantic basin; however, eastern Pacific hurricane activity has increased over the past several years, which means that the western coasts of Mexico and the US could lie in the paths of intensifying hurricanes.
Motivated by Patricia’s extreme nature, Judt and I used the storm as a case study for better understanding hurricane intensification and improving intensity forecasting. We examined the record-breaking intensification period, questioning why the hurricane intensified so rapidly and how it became the most intense hurricane ever observed.
We were able to successfully simulate Patricia’s track and intensity using the Advanced Research Version of the Weather Research and Forecasting (WRF) model, a state-of-the-art numerical weather prediction model that solves discretized versions of the equations governing atmospheric flow. Figure 3 (top) demonstrates that our WRF simulation (red line) predicted Patricia’s storm track with only small deviations from the best estimate of the actual storm’s position and intensity (black line). Even more impressively, the simulation accurately predicted both Patricia’s maximum wind speed and minimum pressure, although the real storm reached those milestones 12 hours earlier than predicted by the WRF.
Resolution was key to the success of our simulation. The Global Forecast System model that we used as a basis for our simulation (see figure 2, red line)—one of the many operational models that drastically underestimated Patricia—featured 15 km resolution. Patricia’s vortex within the model was very coarsely resolved, which is one reason that forecasters never anticipated the storm’s rapid intensification. Our simulation used two vortex-following nested domains featuring 1 km and 3 km resolution. The high resolution allowed our simulation to spin up a better-defined vortex, which behaved similarly to the real storm’s vortex.
We further validated our model’s performance by comparing cross sections of wind speed and sea-level pressure from the simulated storm with Hurricane Hunter aircraft observations (see figure 4). The eye of the hurricane can be seen in the center of the plots, corresponding to zero distance from the storm’s center, with the x-axis extending out 150 km in either direction. Figure 4 shows that the model data compare favorably with the aircraft data and illustrates the canonical structure of an intense hurricane: The annulus of violent winds, the eyewall, surrounds a relatively calm eye featuring the lowest surface pressure.
In the simulated storm, the winds at 3 km altitude in the eyewall reached a maximum speed around 100 m/s, just short of the fastest speed detected for Patricia by aircraft (see top of figure 4). Likewise, the spatial variation of pressure was well-predicted by the WRF simulation, with the modeled storm’s eyewall displaying an exceptional pressure gradient similar to the observations (see bottom of figure 4). The pressure within the eye of the simulated hurricane dipped to 870 mb, close to that of the real storm. The biggest difference was that the eye in our model was some 5 km wider than that measured by the aircraft. The modeled storm also exhibited a smoother behavior than that of the real storm, which would have experienced turbulence and other small-scale wind features.
In addition to improved predictions of hurricane intensity, we hoped our research would help us better understand the environmental conditions that supported Hurricane Patricia’s rapid intensification. In the early 1990s, scientists began compiling data on the factors that contribute to rapid intensification to create the Statistical Hurricane Intensity Prediction Scheme (SHIPS). The researchers found that rapid intensification generally occurs in environments with high sea-surface temperature, low vertical wind shear, and high relative humidity. Due to the explosive nature of Patricia, we expected to discover that all of those variables ranked “extreme” when compared with historical tropical cyclones.
That’s exactly what we found after comparing Patricia with more than 30 years’ worth of eastern Pacific tropical cyclones. Using data from the WRF simulation, we calculated a domain-averaged sea-surface temperature of 30.8 °C for Patricia, which ranks in the 99th percentile historically. The climatological average for storms undergoing rapid intensification is 28.4 °C. The average, however, is based on storms in the Atlantic.
In our model, Patricia experienced a mean vertical wind shear of just 2.5 m/s, lower than all but 15% of storms historically. In fact, the simulated storm briefly faced vertical wind shear below 1 m/s, which is exceedingly low (see top right of figure 5). Finally, we found extremely high moisture content. The historical mean low-level relative humidity for rapidly intensifying hurricanes is 78.6%. The relative humidity of the WRF simulation was above 93%, which ranks Patricia in the 99th percentile. Figure 5 shows the rankings of the WRF simulation’s environmental conditions against historical storm environments.
Our analysis suggests that an extremely favorable environment bolstered Patricia’s rapid intensification. Combined with the higher resolution and vortex initialization of the WRF simulation, the findings offer promise for forecasting efforts, because environmental variables have generally higher predictability than a storm’s internal dynamics.
It is important to note that our research is based on a single WRF simulation that successfully captured Patricia’s intensity evolution; a few of the other simulations we ran did not. Strictly speaking, we cannot yet say whether we just got lucky with the simulation or whether there is some actual predictability. Our next research step is to run a series of simulations with slightly perturbed initial states, to better understand the uncertainty associated with the internal environmental factors influencing the hurricane’s intensity. We hope that teasing out the uncertainty will eventually lead to more reliable hurricane intensity forecasts for future storms as intense as Patricia.
Ryder Fox is an undergraduate student at New Mexico Institute of Mining and Technology. He currently studies hurricane rapid intensification with Falko Judt at the National Center for Atmospheric Research in Boulder, Colorado.
