Hurricane Patricia Reanalysis — Was the Historic Peak Even Stronger? This continues the Category 5 series (Eastern Pacific in this case) with a very thorough analysis of all available data to determine the best single intensity estimate. Official NHC (TCR) — 23 October 2015 Winds: 185 kt (215 mph) Pressure: 872 mb Location: Eastern Pacific (southwest of Manzanillo, Mexico) WxProbe Reanalysis: MSW: 190 kt MSLP: 872 mb Reconnaissance Patricia’s absolute peak intensity occurred during an 11-hour reconnaissance gap. During that interval, the satellite presentation (ADT CI) continued to improve through 1200Z. Here’s the pertinent data: 0446Z AF303: 894 mb raw (center sonde 25 kt → adjusted to 891 mb), max flight-level wind of 179 kt at 4 n mi. 0646Z AF303: 885 mb raw (center sonde treated as 57 kt in TCR → adjusted 879 mb), max flight-level wind of 192 kt at 3 n mi (including 1-second spike of 201 kt). Real-time SFMR of 182 kt; post-processed SFMR peaks to 187 kt (Henning et al. 2016; Klotz reanalysis). RMW contracted to 3 n mi. The aircraft departed while the storm was still deepening. These are the respective observations when recon next sampled the core: 1733Z NOAA3: max flight-level wind was still 192 kt at 8 n mi (VDM; secondary summaries list 191 kt), operational SFMR 166 kt revised to 180 kt after HRD brightness-temperature reprocessing, pressure already higher (VDM at 883 mb with 45 kt → adjusted to 878 mb). Early ERC signatures present. 2044Z: 910 mb with 68 kt → adjusted to 903 mb; flight-level wind of 160 kt. Rapid filling well underway. The observed rates while aircraft were present taking into consideration the estimated lowest pressure of 872 mb at 1200Z: * 0446Z–0646Z (2 h): pressure −6 mb/hr, flight-level wind +6.5 kt/hr, RMW 4 to 3 n mi. * 1200Z–1733Z (5.55 h): pressure +1.1 mb/hr (872 → 878 mb). * 1733Z–2044Z (3.2 h): 8 mb/hr filling rate once ERC matured. Pressure Adjusted pressure at the 192 kt pass was 879 mb. The satellite organization continued to improve until 1200Z. A moderated continuation of the pre-gap −6 mb/hr rate (1.3 mb/hr, which is 22% of the observed rate) across the 5.2 hour gap produces 872 mb at 1200Z. This connects directly to the next observation: 872 mb → 878 mb at 1733Z = +1.1 mb/hr, matching the precisely calculated slow initial filling rate. Anything deeper than 872 mb is possible, but not assured. Structure As shown above, the RMW contracted to an astoundingly small 3 n mi at the 192 kt maximum (pinhole). In addition, mean of the nonzero R34 was compact at 110 n mi, with hurricane-force winds only to 15–16 n mi from the center. Single wall at peak. No size veto required. Reduction Factor A radius of maximum wind of ≤ 5 n mi and pressure below 900 mb support the 0.95 bin (Powell et al. 2009; Patricia SFMR-to-flight-level ratios of 0.95). Patricia’s values were obviously well-below those thresholds. Flight-level interpolation (single undersample step): Pre-gap rate was +6.5 kt/hr. A continuation of *only* +1.2 kt/hr (18% of the observed rate) for 5.2 hours = +6 kt. This translates as 192 kt → 198 kt at 1200Z. Return leg: 198 kt → 192 kt at 1733Z = −1.1 k/hr, consistent with the +1.1 mb/h slow filling while the flight-level wind remained at 192 kt. These equate to 198 kt × 0.95 = 188 kt → 190 kt, constrained by the measured SFMR values already in the 180–187 kt range. The undersample step is counted once via the interpolation, and 198 kt is the conservative end of that single step. I’ll add that the 192 kt /187 kt pair (ratio of 0.97) and other observed SFMR-to-flight-level ratios near 0.94–0.95 indicate that a higher combination — 202 kt flight-level with a 0.96 factor — would yield 195 kt. This remains a possible high-end, but is not assured. Without direct sampling in the peak window, it isn’t an advisable choice under the methodology, which caps the factor at 0.95 and publishes the conservative single-step result. Pressure Check KCZ on 872 mb and the compact R34 is consistent with the upper end of the observed range. Final Estimate: MSW: 190 kt (220 mph) MSLP: 872 mb RMW: 3 n mi Time: 1200 UTC 23 October 2015 That is a 5 kt revision of the official wind, while pressure remains unchanged. This further solidifies Patricia’s place as containing the most extreme intensity ever reliably measured in a tropical cyclone, worldwide. Framework Notes The flight-level-to-surface factor is selected by the radius of the sampled wind maximum, with pressure and outer size able to veto that first guess. In the most compact cores (RMW ≤ 5 n mi and pressure under 900 mb), the factor can reach 0.95. This ceiling is supported by Powell et al. (2009) and by the Patricia 2015 observations themselves (Rogers et al. 2017; Henning et al. 2016; NHC TCR), where SFMR-to-flight-level ratios reached 0.95. The operational mean remains at 0.90 (Franklin et al. 2003). A small RMW does not justify the highest factors if the mean of the nonzero R34 quadrants is on the order of 140 n mi, or an outer eyewall is active. An undersample of peak intensity is counted once only. Knaff–Zehr–Courtney (KZC) is used as a consistency check. The hour is assigned to the locked apex pair. Sources: NHC TCR (Kimberlain et al. 2016); Henning et al. (2016) AMS preprint; Rogers et al. (2017) BAMS; HRD SFMR reprocessing (Klotz); vortex data messages 0446Z, 0646Z, 1733Z, 2033Z.

Two quick points: 1) I agree that the physics-based models are still valuable and 2) no singular model will ever be infallible. That said, there wasn’t *that* much difference in the track errors between the two (physics vs AI) over the past seven days. It is just magnified simply because #Champi made an unpredictable dramatic wobble towards land. Even if the disparity in this case were greater, it wouldn’t override the overwhelming superiority in track performance of the AI models throughout this season. But again, I still find value in the physics-based guidance as well. In short, they each have their relative strengths and weaknesses, and that should always be taken into consideration when producing a forecast. The NHC has recently been weighting GDMI a little too heavily right now. Recent East Pacific discussions make that clear: the official track is repeatedly and consistently described as lying closest to, closely follows, or in best overall agreement with, the Google DeepMind ensemble mean and HCCA. In practice, the forecast has often mirrored that guidance rather than fully synthesizing the suite. GDMI has earned significant weight through strong track performance this season, and physics-based models remain imperfect. But every system still has their relative strengths and failure modes. When the official forecast tracks the leading AI guidance this closely, it also inherits those biases rather than being checked by the dynamical models that still handle certain processes (terrain interaction, intensity evolution, specific steering regimes) better in individual cases. Consequently, I feel no single model — or heavily weighted pair — should necessarily dominate the process.

One year ago this morning, NHC Discussion #8 and the concurrent WxProbe update were issued on the still heavily-sheared Tropical Storm #Melissa. Here’s what each of us were expecting: Forecasts valid 1800Z 26 & 27 October 2025: NHC (interpolated from 23 Oct): 26/1800Z — 16.8N 76.5W/ 95 kt 27/1800Z — 17.0N 77.5W/ 110 kt WxProbe (23 Oct): 26/1800Z — 16.8N 77.7W/ 130 kt 27/1800Z — 17.1N 79.2W/ 145 kt Actual (NHC best track): 26/1800Z — 16.3N 76.9W/ 120 kt 27/1800Z — 16.4N 78.3W/ 155 kt The verified track essentially split the difference: NHC was too slow and too far east, while the WxProbe forecast was too fast and too far west. Intensity tracked much closer to, and very near, the WxProbe numbers. That stronger-and-more-west versus more-conservative-and-more-east contrast was a consistent theme in the respective forecasts throughout the event. #Hurricane Melissa first reached Category 5 intensity at 0900Z on the 27th — precisely in the window predicted in that October 23 update — while the contemporaneous NHC forecast from that morning still only projected 105 kt (Category 3) at that same hour and had only 115 kt a full day later. Even as late as 1500Z on the 25th (Discussion 17), NHC was forecasting a peak of 135 kt and explicitly stated they were not yet ready to call for a Category 5, citing ongoing track uncertainty. By the 2100Z advisory that evening (Discussion 18), they had finally raised the forecast peak to the 140 kt Cat 5 threshold. Expectations of earlier improving environmental conditions and the development of a well-defined inner core allowed for the stronger solution. That much earlier call for rapid intensification south of Jamaica once conditions became more favorable proved directionally correct on both timing and magnitude, even if the precise longitude still needed refinement at the 72–96 hour range. The NHC’s more easterly and slower-intensifying solution underestimated both the rate and the ultimate magnitude of strengthening once the shear relaxed, remaining 25–45 kt too low at the 26–27/1800Z times and still only calling for a 115 kt peak a full day after Melissa had already reached Category 5 strength. This is an example of where the NHC intensity forecasts have historically shown a conservative bias in some rapid-intensification cases when track or shear uncertainty remains elevated. Specific WxProbe forecast from compromised/inaccessible account: xitter.cf/Mr_XMusk27/status/1981…

Independent Reanalysis of Hurricane Katrina’s Peak Intensity This continues the Atlantic Category 5 #Hurricane series. Peak only. Landfall is a separate post. Official NHC (TCR) — 28 August 2005, 1800 UTC Winds: 150 kt (175 mph) Pressure: 902 mb Eye: 22 n mi at the 166 kt flight-level maximum WxProbe Reanalysis: MSW: 155 kt MSLP: 902 mb Reconnaissance AF302 measured a 700 mb flight-level wind of 166 kt in the northeast eyewall at 1422Z. A higher wind could have occurred in the aircraft gap. The last AF302 center fix was at 1500Z, with a pressure of 907 mb. The first NOAA3 fix was at 1755Z, which measured 902 mb. Nothing was flown in the eyewall for 175 minutes while the pressure fell 5 mb. The core was expanding across that gap: the eye was 22 n mi before it, 25 n mi at the 902 mb fix, and 28 n mi by 1923Z, and the radius of maximum wind had moved from 14–18 n mi out to 22 n mi. Given that expansion, the peak wind is treated as not missed, and the 166 kt remains the best reading. Pressure As mentioned above, the minimum was measured at 1755Z and again at 1923Z, 88 minutes later. The next fix was 903 mb, followed by 904 mb. The only hole is 1500Z to 1755Z. A transient value below 902 mb during that gap is possible, but not assured. Consequently, the official 902 mb is retained. Structure The RMW was 17 n mi from the center at the time of the 166 kt peak flight-level wind, which matched the same-mission cluster of 16–18 n mi. The mean of the nonzero R34 quadrants is 165 n mi at the peak hour. The 17 n mi inner core doesn’t erase that relatively large outer field. Reduction Factor RMW: 17 n mi Pressure: 902 mb R34: 165 n mi The top bins (0.94, 0.95) are reserved for small-core, compact hurricanes with a central pressure <910 mb. Katrina meets the requirements on pressure, but not on size of its wind field. The 17 n mi wall is the 0.93 “width” path. The 165 n mi mean R34 is the reason the factor sits just below that, at 0.925, rather than on it. The factor used is 0.925. 166 × 0.925 = 154 kt, which rounds to 155. Pressure Check CKZ, mean R34, 26.0–26.3°N, translation 10–11 kt, environmental pressure assumed 1010–1012 mb: 154–156 kt at 902 mb. In the equation, S is 0.80–0.89, so the relatively large wind field is accounted for in the solution. Final Estimate MSW: 155 kt (180 mph) MSLP: 902 mb RMW: 17 n mi Time: 1700 UTC 28 August 2005 That is a 5 kt revision of the official wind, no revision of official pressure, and a one-hour shift of the peak from the 1800Z pressure hour to the hour assigned to the pair. Series Comparison Wilma and the 1935 Labor Day Hurricane remain 170 kt. Melissa is 165 kt. Katrina ties Andrew and Camille at 155 kt, with a larger RMW and a much larger R34 than either.

Looking Back on Hurricane Isaias — One Week Later Hurricane #Isaias peaked as a Category 3 major hurricane (105 kt/120 mph, 959 mb) on October 9 as it was closing in on the Florida Panhandle. It weakened thereafter due to increasing wind shear and dry air entrainment. The center made landfall near Destin, Florida, at about 8:30 p.m. CDT on October 9 as a Category 2 (90 kt/105 mph, 971 mb). The leading northern edge of the eyewall reached the Okaloosa Island/Fort Walton Beach area while the storm was still a legitimate Category 3 (100 kt); it then weakened by roughly 10 kt over the final 2.5 hours before the center actually crossed the coast. On a personal note, I was positioned on the narrow and exposed road on Okaloosa Island between Fort Walton Beach and Destin, where wind gusts exceeded 100 mph as the western and southwestern parts of the eyewall moved through. This constituted the 46th U.S. mainland #hurricane eyewall intercept and the 19th major (as it was still a Cat 3 when the aforementioned northern edge moved overhead). The storm produced an unusually broad and asymmetric wind footprint for its official Category 2 landfall intensity. Recorded gusts of 100+ mph occurred from the Pensacola City Hall area to Freeport (~55 nmi along the coast), including within the eyewall and extending well outside it in the western quadrant. In addition, hurricane-force gusts (74+ mph) spanned from Orange Beach, Alabama, to Beacon Hill just east of Mexico Beach (~120 nmi). The peak measured gusts reached 116–118 mph near Oriole Beach and Jay. A sting-jet-like feature on the western edge of the precipitation shield (during the onset of extratropical transition) helped drive the unusually strong western-quadrant winds. These winds caused widespread tree and power-line damage, roof losses on beachfront hotels and other structures, and collapsed gas-station canopies. More than 900,000 customers lost power across Florida, Alabama, and Georgia (peak reports were around 910,000–911,000). Worst of all, at least nine fatalities were confirmed (six in Florida, two in Alabama, and one in South Carolina), mostly from falling trees and related incidents. Storm-surge inundation of roughly 5.6 ft was observed at Panama City Beach. A Flash Flood Emergency was issued for southern Okaloosa County due to torrential rainfall, with widespread 5–8 inch totals and localized amounts exceeding 12 (including over 11 inches near Mary Esther and Fort Walton Beach), producing major freshwater flooding that caused significant damage to homes and businesses in that area. Preliminary estimates of losses surpass $2 billion. For residents of the western Florida Panhandle, Isaias is a storm that will be talked about for generations to come.

Jeff here is arguing that the recent backloaded #hurricane seasons (quiet peaks, late activity, and more Florida Gulf hits) are mainly a climate-change stability signal from warmer summers. The same period is better explained by the MJO. The predominant background state has been a warm AMO since 1995, with ENSO as the main seasonal modulator. On top of that, the MJO has been the primary subseasonal control. In 2024, 2025, and this season, the MJO was unfavorable or weak through most of the peak (until after 9/16), then shifted more favorable later — exactly when activity increased. The 2023 season had the opposite alignment and produced a hyperactive peak instead. Year-to-year timing differences track the MJO, *not* a uniform stability suppression. The MJO has been the dominant reason for the quiet peaks and late bursts in these years.

Hurricane #Isaias delivered a very impressive wind footprint —especially for a Category 2 landfall — producing *recorded* gusts of 100+ mph from Pensacola City Hall to Freeport (~55 n mi along the coast), including within the eyewall itself and extending well outside of it in the western quadrant. Hurricane-force gusts (74+ mph) spanned from Orange Beach, AL to Beacon Hill — located just east of Mexico Beach (~120 nmi). This broad and asymmetric wind field featured unusually strong winds in the western quadrant, driven by a sting-jet-like feature on the western edge of the precipitation shield as the storm began extratropical transition. As seen in the attached video, my intercept location endured the peak winds in the western and southwestern parts of the eyewall, with wind gusts well over 100 mph as well.

Before Isaias — There was Michael Today marks the eighth anniversary of the last Category 5 #hurricane to strike the U.S. mainland. Hurricane #Michael claimed more than 50 lives and forever changed countless more. Sadly, many lost everything they owned—including homes their families had maintained for generations. The hardest-hit town of Mexico Beach was nearly wiped off the map. Although it has slowly recovered, its trademark “forgotten coast” character was largely erased. The combination of its extreme intensity at landfall and relatively fast forward movement allowed it to devastate communities hundreds of miles inland. Total damage exceeded $25 billion. I was on Mexico Beach that fateful afternoon documenting the landfall firsthand. I recorded the absolute peak—gusts over 200 mph and 14–15’ of storm surge—as the eyewall tore through and destroyed the town. Here’s what I saw. The brunt of the winds runs roughly from the 58:00 mark to about 1:18:00 (note that the footage from 1:04:00 to 1:07:00 is from much earlier in the eyewall, preceding the peak): youtu.be/wiL3QKraaTY