WEBVTT 00:00:00.000 --> 00:00:03.140 If you felt like 2025 was a year where every 00:00:03.140 --> 00:00:07.000 conversation, every headline, and maybe even 00:00:07.000 --> 00:00:09.019 your insurance premium was just dominated by 00:00:09.019 --> 00:00:11.939 extreme weather, you weren't wrong. Not at all. 00:00:12.160 --> 00:00:13.900 It felt like weather wasn't just something we 00:00:13.900 --> 00:00:16.219 predicted anymore. It was, I don't know, it was 00:00:16.219 --> 00:00:19.690 this persistent, unavoidable reality. that was 00:00:19.690 --> 00:00:22.129 just reshaping everything. I think that's a perfect 00:00:22.129 --> 00:00:24.010 way to put it. It was a year that really forced 00:00:24.010 --> 00:00:26.809 a shift in perspective. And what's crucial for 00:00:26.809 --> 00:00:28.910 us to get our heads around right away is that 00:00:28.910 --> 00:00:31.949 despite natural processes that should have acted 00:00:31.949 --> 00:00:34.329 as a kind of break on global heat. Right, these 00:00:34.329 --> 00:00:36.530 natural cooling cycles. Exactly, cycles we've 00:00:36.530 --> 00:00:39.450 historically relied upon. Despite those, global 00:00:39.450 --> 00:00:42.070 heat records still remained wildly persistently 00:00:42.070 --> 00:00:43.990 above the baseline we've known. So we're looking 00:00:43.990 --> 00:00:47.259 back at a year of Well, undeniable severity. 00:00:47.579 --> 00:00:50.340 Undeniable severity, but also, and this is critical, 00:00:50.939 --> 00:00:54.259 a year of these unprecedented technological leaps 00:00:54.259 --> 00:00:56.820 in how we actually track these relentless extremes. 00:00:57.119 --> 00:00:59.740 Welcome to Meteorology Matters, the podcast that 00:00:59.740 --> 00:01:02.259 dives deep into the science, chaos, and stories 00:01:02.259 --> 00:01:04.700 behind the weather that shapes our world. This 00:01:04.700 --> 00:01:07.019 show was created by meteorologist Rob Jones. 00:01:07.260 --> 00:01:10.040 Now let's get into today's episode of Meteorology 00:01:10.040 --> 00:01:12.579 Matters. So let's start this deep exploration 00:01:12.579 --> 00:01:15.599 by looking back at the sheer scale of what happened 00:01:15.599 --> 00:01:19.140 globally in 2025. It's the universal question 00:01:19.140 --> 00:01:21.379 at the end of every year, isn't it? Was it a 00:01:21.379 --> 00:01:24.079 bad year for extreme weather? It is. And based 00:01:24.079 --> 00:01:25.939 on all the research, all the analysis we've seen, 00:01:26.079 --> 00:01:29.640 the answer is tragically. an unequivocal yes 00:01:29.640 --> 00:01:31.799 again. That yes feels almost too simple. Oh, 00:01:31.819 --> 00:01:34.140 it's deceptive in its simplicity because it needs 00:01:34.140 --> 00:01:37.280 so much context. When researchers finally tallied 00:01:37.280 --> 00:01:41.079 the numbers, 2025 was, believe it or not, globally 00:01:41.079 --> 00:01:43.859 slightly cooler than the record -shattering year 00:01:43.859 --> 00:01:47.480 of 2024. It's a tiny dip. Tiny dip? But here's 00:01:47.480 --> 00:01:49.340 the critical point, the one that tells you everything 00:01:49.340 --> 00:01:51.659 about the state of our climate. Despite that 00:01:51.659 --> 00:01:54.739 dip, 2025 was still far, far hotter than almost 00:01:54.739 --> 00:01:58.299 any other year on record before 2023. That fact 00:01:58.299 --> 00:02:01.000 alone just confirms we are operating from a dramatically 00:02:01.000 --> 00:02:04.680 new baseline. That slight cooling feels like 00:02:04.680 --> 00:02:07.920 a, well, a meteorological contradiction. If the 00:02:07.920 --> 00:02:09.900 Earth takes a tiny step back on temperature, 00:02:10.219 --> 00:02:12.099 why didn't they give us more widespread relief? 00:02:12.479 --> 00:02:14.599 Because we're dealing with this massive underlying 00:02:14.599 --> 00:02:17.639 human -driven force that is just overwhelming 00:02:17.639 --> 00:02:21.419 natural variability. The really surprising detail, 00:02:21.680 --> 00:02:24.219 meteorologically speaking, is that this persistent 00:02:24.219 --> 00:02:27.159 extreme heat happened even while the Pacific 00:02:27.159 --> 00:02:29.919 was experiencing weak La Nina conditions. OK, 00:02:29.919 --> 00:02:32.180 so let's just quickly clarify La Nina for everyone 00:02:32.180 --> 00:02:34.259 listening. It's the opposite of El Nino. Right. 00:02:34.340 --> 00:02:37.620 El Nino is this recurring natural thing, a phenomenon 00:02:37.620 --> 00:02:41.240 characterized by unusually cool sea surface temperatures 00:02:41.240 --> 00:02:44.099 across the central and eastern equatorial Pacific. 00:02:44.180 --> 00:02:46.680 And that has a big impact. A huge atmospheric 00:02:46.680 --> 00:02:49.689 impact. Historically, that cooling effect tends 00:02:49.689 --> 00:02:52.870 to pull the average global temperature down slightly 00:02:52.870 --> 00:02:55.129 for that year. It's one of nature's biggest breaks 00:02:55.129 --> 00:02:58.250 on global heating. So the fact that 2025 was 00:02:58.250 --> 00:03:01.330 still historically incredibly hot, even with 00:03:01.330 --> 00:03:03.689 La Niña trying to hit the brakes, that means 00:03:03.689 --> 00:03:06.449 the human driven warming signal is now so strong 00:03:06.449 --> 00:03:08.610 it could just neutralize one of nature's biggest 00:03:08.610 --> 00:03:11.770 cooling systems. Exactly. The sheer strength 00:03:11.770 --> 00:03:14.729 of anthropogenic warming, that's the warming 00:03:14.729 --> 00:03:17.569 caused by human activity, is just laid bare by 00:03:17.569 --> 00:03:20.039 this paradox. It means the heat that drove all 00:03:20.039 --> 00:03:22.240 those destructive extremes across every continent 00:03:22.240 --> 00:03:24.500 wasn't just some random spike. It was the new 00:03:24.500 --> 00:03:26.780 normal. It was the climate system operating at 00:03:26.780 --> 00:03:29.500 a dramatically elevated level. This isn't some 00:03:29.500 --> 00:03:32.759 distant theoretical threat anymore. It is our 00:03:32.759 --> 00:03:35.979 present undeniable reality. And the scale of 00:03:35.979 --> 00:03:39.039 the impact really supports that conclusion. I 00:03:39.039 --> 00:03:40.780 mean, researchers didn't just look at temperature 00:03:40.780 --> 00:03:44.039 anomalies. They analyzed 22 of the most significant 00:03:44.039 --> 00:03:47.000 extreme weather events globally. A terrible mosaic, 00:03:47.300 --> 00:03:49.800 really? It really was. Catastrophic heatwaves, 00:03:50.199 --> 00:03:52.900 intense floods, devastating storms, persistent 00:03:52.900 --> 00:03:55.400 droughts, and these uncontrollable wildfires 00:03:55.400 --> 00:03:57.819 that just claimed countless lives, destroyed 00:03:57.819 --> 00:04:00.240 infrastructure, and crippled communities. And 00:04:00.240 --> 00:04:02.939 every story embedded in those events is tied 00:04:02.939 --> 00:04:04.759 directly to something that sounds clinically 00:04:04.759 --> 00:04:07.409 small. It's called incremental warming. We're 00:04:07.409 --> 00:04:10.150 talking about that slow, steady, upward creep 00:04:10.150 --> 00:04:12.569 of global average temperatures, fractions of 00:04:12.569 --> 00:04:15.090 a degree, that have these disproportionately 00:04:15.090 --> 00:04:18.189 massive real -world impacts. To put that in perspective 00:04:18.189 --> 00:04:21.029 for everyone, we often talk about the 2015 Paris 00:04:21.029 --> 00:04:24.089 Agreement as this kind of benchmark moment. So 00:04:24.089 --> 00:04:26.970 how much warming have we actually seen just in 00:04:26.970 --> 00:04:28.550 the decades since that agreement was signed? 00:04:28.750 --> 00:04:31.930 Since 2015, global warming has increased by 0 00:04:31.930 --> 00:04:36.449 .3 degrees Celsius. 0 .3. Just sounds so tiny. 00:04:36.629 --> 00:04:39.250 It sounds tiny. It sounds scientifically insignificant. 00:04:39.769 --> 00:04:42.569 But that fraction represents a massive escalation 00:04:42.569 --> 00:04:45.009 of risk that has fundamentally changed the odds 00:04:45.009 --> 00:04:48.290 of extreme weather. That number, .3 degrees C, 00:04:48.449 --> 00:04:51.120 is so abstract to most people. What does that 00:04:51.120 --> 00:04:54.259 tiny incremental warming actually mean in terms 00:04:54.259 --> 00:04:55.819 of daily life, you know, for you and me just 00:04:55.819 --> 00:04:57.480 walking outside? It's the difference between 00:04:57.480 --> 00:05:00.699 a harsh summer and a catastrophic one. That seemingly 00:05:00.699 --> 00:05:03.420 small rise has already resulted in extreme heat 00:05:03.420 --> 00:05:05.839 becoming significantly more frequent across the 00:05:05.839 --> 00:05:09.120 entire globe, adding on average 11 extra hot 00:05:09.120 --> 00:05:11.860 days per year. 11 extra days? Think about that. 00:05:11.980 --> 00:05:15.180 That's nearly two full weeks of additional potentially 00:05:15.180 --> 00:05:17.540 dangerous, life -threatening heat every single 00:05:17.540 --> 00:05:20.269 year for the whole planet. 11 extra hot days 00:05:20.269 --> 00:05:23.399 on average. That statistic doesn't just sound 00:05:23.399 --> 00:05:26.100 chilling it. I mean, it means a fundamental restructuring 00:05:26.100 --> 00:05:28.779 of what a normal year even looks like. It does. 00:05:28.779 --> 00:05:31.180 And it means the type of heat is changing too. 00:05:31.259 --> 00:05:33.980 It's not just the dry, you know, high temperature 00:05:33.980 --> 00:05:37.180 heat we worry about. It's the increasingly frequent 00:05:37.180 --> 00:05:39.860 wet bulb heat events. Right. Where the humidity 00:05:39.860 --> 00:05:42.860 is the real killer. That's the one. Wet bulb 00:05:42.860 --> 00:05:45.220 temperature measures heat and humidity combined. 00:05:45.459 --> 00:05:48.040 And when it rises past a certain point, the human 00:05:48.040 --> 00:05:50.500 body literally cannot cool itself by sweating. 00:05:51.279 --> 00:05:54.399 An increase of 0 .3 degrees C pushes regions, 00:05:54.839 --> 00:05:57.180 especially near the equator, closer and closer 00:05:57.180 --> 00:06:00.360 to those critical, lethal, wet bulb thresholds. 00:06:00.600 --> 00:06:03.279 So those 11 extra days are just much more dangerous 00:06:03.279 --> 00:06:04.980 than they would have been in the past. Far more 00:06:04.980 --> 00:06:07.720 dangerous. And the effects just ripple out from 00:06:07.720 --> 00:06:10.079 there. Agriculture is strained because crops 00:06:10.079 --> 00:06:12.180 have specific temperature tolerances. You see 00:06:12.180 --> 00:06:15.060 mass crop failures, reduced yield. And our infrastructure. 00:06:15.500 --> 00:06:17.720 Our infrastructure, which was designed for a 00:06:17.720 --> 00:06:19.579 climate that doesn't exist anymore, just starts 00:06:19.579 --> 00:06:22.870 to buckle. You have asphalt roads melting, rail 00:06:22.870 --> 00:06:25.990 lines warping, electricity grids struggling because 00:06:25.990 --> 00:06:28.769 the demand for cooling spikes exponentially with 00:06:28.769 --> 00:06:32.110 just a few extra degrees. So the warming makes 00:06:32.110 --> 00:06:34.829 heat waves longer and they happen more often. 00:06:34.970 --> 00:06:37.370 So there's no time to recover. Exactly. Nature 00:06:37.370 --> 00:06:39.870 can't recover. Human systems can't recover. There's 00:06:39.870 --> 00:06:42.870 no break. And the change in probability for specific 00:06:43.050 --> 00:06:45.769 Local events is even more alarming than that 00:06:45.769 --> 00:06:48.389 global average. Oh, absolutely. Okay for specific 00:06:48.389 --> 00:06:50.589 well -documented heat waves We've seen in previous 00:06:50.589 --> 00:06:53.300 years like that extreme heat that devastated 00:06:53.300 --> 00:06:55.959 parts of the Amazon or the severe events that 00:06:55.959 --> 00:06:58.480 hit West African nations like Burkina Faso and 00:06:58.480 --> 00:07:00.980 Mali. Researchers found that the probability 00:07:00.980 --> 00:07:03.399 of the specific high intensity events happening 00:07:03.399 --> 00:07:06.579 again has increased almost 10 times since 2015. 00:07:06.720 --> 00:07:09.399 10 times more likely. 10 times. Every single 00:07:09.399 --> 00:07:11.279 fraction of a degree matters because it shifts 00:07:11.279 --> 00:07:14.490 the odds exponentially toward disaster. It makes 00:07:14.490 --> 00:07:16.949 events that were once rare anomalies, you know, 00:07:17.089 --> 00:07:19.670 once in a century events, now disturbingly routine. 00:07:19.970 --> 00:07:22.990 That analysis just drives home the urgency for, 00:07:22.990 --> 00:07:25.930 well, for both mitigation and adaptation. We're 00:07:25.930 --> 00:07:28.370 getting very close to the limits of what society 00:07:28.370 --> 00:07:30.790 can actually absorb. And when we talk about who 00:07:30.790 --> 00:07:34.069 absorbs that cost, we have to look beyond the 00:07:34.069 --> 00:07:36.660 generalized impact. Extreme weather does not 00:07:36.660 --> 00:07:39.759 distribute its effects equally. No, it consistently 00:07:39.759 --> 00:07:42.680 and devastatingly hits those who are already 00:07:42.680 --> 00:07:45.000 marginalized and vulnerable, the hardest. We 00:07:45.000 --> 00:07:47.399 saw some particularly detailed research that 00:07:47.399 --> 00:07:50.660 illustrated this profound inequity across regions 00:07:50.660 --> 00:07:53.620 of the global south in 2025. Yes, there was a 00:07:53.620 --> 00:07:55.639 study focused on South Sudan, which provided 00:07:55.639 --> 00:07:58.939 a powerful, really granular example of how these 00:07:58.939 --> 00:08:01.579 climate hazards intersect with preexisting social 00:08:01.579 --> 00:08:03.620 structures. What do they find? They found that 00:08:03.620 --> 00:08:06.110 women in South Sudan are disproportionately affected 00:08:06.110 --> 00:08:09.189 by extreme heat, mainly because of their concentration 00:08:09.189 --> 00:08:11.689 in informal heat exposed work. Can you paint 00:08:11.689 --> 00:08:13.769 a clearer picture of what that informal work 00:08:13.769 --> 00:08:15.509 actually looks like on the ground? We're talking 00:08:15.509 --> 00:08:17.649 about essential economic activity that often 00:08:17.649 --> 00:08:21.069 provides zero shelter. Things like subsistence 00:08:21.069 --> 00:08:23.850 agriculture, spending hours tending fields under 00:08:23.850 --> 00:08:26.930 intense sun, or jobs like street vending, where 00:08:26.930 --> 00:08:29.769 the only place to do business is directly exposed 00:08:29.769 --> 00:08:32.529 to peak daytime heat. And these are jobs that 00:08:32.529 --> 00:08:35.139 are critical for family. Absolutely critical, 00:08:35.200 --> 00:08:38.100 but they offer zero mitigation against severe 00:08:38.100 --> 00:08:40.840 heat. So it's not just the exposure to the heat 00:08:40.840 --> 00:08:44.100 itself, and during a heat wave, that work gets 00:08:44.100 --> 00:08:47.120 even harder. It intensifies. Family members are 00:08:47.120 --> 00:08:49.379 more likely to fall ill, so their own exposure 00:08:49.379 --> 00:08:51.539 to dangerous conditions increases while their 00:08:51.539 --> 00:08:54.000 mobility and ability to earn a living shrinks. 00:08:54.639 --> 00:08:56.879 And the impact spills into the next generation. 00:08:57.639 --> 00:09:01.179 Extreme heat forces, widespread school closures, 00:09:01.519 --> 00:09:03.480 which immediately interrupts the education of 00:09:03.480 --> 00:09:09.740 girls more severely than boys. perpetuating the 00:09:09.740 --> 00:09:12.279 cycle of vulnerability. Exactly. What's equally 00:09:12.279 --> 00:09:14.399 important and something the analysis really pointed 00:09:14.399 --> 00:09:17.159 out is that this inequity extends right into 00:09:17.159 --> 00:09:19.580 the scientific process itself. It's not just 00:09:19.580 --> 00:09:21.860 that the impact is unequal. It's the evidence 00:09:21.860 --> 00:09:24.600 we have to study those impacts. It creates a 00:09:24.600 --> 00:09:26.919 kind of scientific blind spot. That is a critical 00:09:26.919 --> 00:09:29.580 finding. Many studies that focused on severe 00:09:29.580 --> 00:09:32.559 weather in the global south in 2025, specifically 00:09:32.559 --> 00:09:35.220 heavy rainfall, they ran into the same problem 00:09:35.220 --> 00:09:38.019 over and over again. Which one? Significant gaps. 00:09:38.090 --> 00:09:40.970 in localized on -the -ground observational data. 00:09:41.549 --> 00:09:44.049 And even worse, the studies had to rely heavily 00:09:44.049 --> 00:09:46.909 on global climate models that were fundamentally 00:09:46.909 --> 00:09:50.309 developed, calibrated, and optimized, primarily 00:09:50.309 --> 00:09:52.190 for the conditions and infrastructure of the 00:09:52.190 --> 00:09:54.389 global north. So we're trying to understand the 00:09:54.389 --> 00:09:56.710 hyperlocal effects of massive rainfall in a place 00:09:56.710 --> 00:09:59.570 like South Sudan using models that were optimized 00:09:59.570 --> 00:10:02.750 for, what, Kansas or Germany? In effect, yes. 00:10:03.090 --> 00:10:06.480 The scientific foundation is uneven. When you 00:10:06.480 --> 00:10:09.419 lack that granular, reliable local data, and 00:10:09.419 --> 00:10:11.220 your predictive tools are biased toward different 00:10:11.220 --> 00:10:13.820 atmospheric conditions and terrains, it severely 00:10:13.820 --> 00:10:16.440 hinders your ability to draw confident, actionable 00:10:16.440 --> 00:10:18.820 conclusions. And that cripples local planning 00:10:18.820 --> 00:10:20.960 where it's needed most. It absolutely cripples 00:10:20.960 --> 00:10:23.320 it, in the regions already experiencing the most 00:10:23.320 --> 00:10:25.860 immediate and severe consequences. Addressing 00:10:25.860 --> 00:10:28.100 this is just as vital as reducing emissions. 00:10:28.580 --> 00:10:30.940 We need investment in closing these scientific 00:10:30.940 --> 00:10:33.100 data gaps globally so marginalized communities 00:10:33.100 --> 00:10:35.639 can build genuinely effective data -driven resilience. 00:10:36.159 --> 00:10:38.899 That sets a very intense, very sobering stage 00:10:38.899 --> 00:10:42.779 for 2025. The overall warming picture is relentless 00:10:42.779 --> 00:10:46.080 and the human cost is so unequally distributed. 00:10:46.700 --> 00:10:48.980 So let's pivot now from that persistent global 00:10:48.980 --> 00:10:53.480 heat to a specific terrifying example of catastrophic 00:10:53.480 --> 00:10:56.600 weather. the Atlantic hurricane season. This 00:10:56.600 --> 00:10:59.000 was a year where the early forecast predicting 00:10:59.000 --> 00:11:03.039 a highly active season, well, they proved devastatingly 00:11:03.039 --> 00:11:05.779 accurate. The season was undeniably busy. Overall, 00:11:06.019 --> 00:11:09.240 we saw 13 named storms, which is a busy count, 00:11:09.700 --> 00:11:12.019 and frighteningly, three of those intensified 00:11:12.019 --> 00:11:14.460 into category five hurricanes. Three cat fives? 00:11:14.620 --> 00:11:17.639 Three. This puts 2025 among the most intense 00:11:17.639 --> 00:11:20.100 seasons on record, just based on that peak strength. 00:11:20.320 --> 00:11:22.100 There was one small silver lining, for the U 00:11:22.100 --> 00:11:23.940 .S. coastline at least. For the first time in 00:11:23.940 --> 00:11:26.139 a decade, a hurricane did not make landfall on 00:11:26.139 --> 00:11:29.019 American soil. That localized good fortune unfortunately 00:11:29.019 --> 00:11:30.779 did not extend to our neighbors in the Caribbean. 00:11:31.080 --> 00:11:32.980 We have to zero in on the most destructive storm 00:11:32.980 --> 00:11:35.899 of the year, Hurricane Melissa. A truly devastating 00:11:35.899 --> 00:11:38.759 storm. It was a monster, a category five that 00:11:38.759 --> 00:11:41.720 achieved maximum intensity, slamming Jamaica 00:11:41.720 --> 00:11:46.080 in late October with sustained winds of 185 miles 00:11:46.080 --> 00:11:50.059 per hour. It devastated communities and, tragically, 00:11:50.340 --> 00:11:52.980 resulted in dozens of deaths. Hurricane Melissa, 00:11:53.279 --> 00:11:56.080 in its sheer intensity, it really highlighted 00:11:56.080 --> 00:11:58.220 a painful lesson that ties right back to our 00:11:58.220 --> 00:12:00.820 discussion on adaptation in this rapidly warming 00:12:00.820 --> 00:12:03.440 world. It showed the clear limits of preparedness. 00:12:03.960 --> 00:12:06.220 We talk constantly about the need for adaptation 00:12:06.220 --> 00:12:09.019 building better seawalls, improving warning systems 00:12:09.019 --> 00:12:11.580 and forcing building codes, and those are all 00:12:11.580 --> 00:12:14.179 absolutely essential for preventing massive loss 00:12:14.179 --> 00:12:16.919 of life from less intense storms. But Melissa 00:12:16.919 --> 00:12:19.139 was different. Melissa made it painfully clear. 00:12:19.320 --> 00:12:21.500 When a storm of that maximum intensity strikes 00:12:21.500 --> 00:12:24.539 a small island nation like Jamaica, even relatively 00:12:24.539 --> 00:12:27.259 high levels of preparedness simply cannot prevent 00:12:27.259 --> 00:12:29.940 catastrophic losses and infrastructure annihilation. 00:12:30.360 --> 00:12:33.480 The physical force is just too great. Which emphasizes 00:12:33.480 --> 00:12:36.259 that forecasting alone isn't the solution, but 00:12:36.259 --> 00:12:38.460 it is that critical first step. And this is where 00:12:38.460 --> 00:12:40.519 the story gets really compelling, shifting from 00:12:40.519 --> 00:12:42.539 the tragedy of the storm to the technological 00:12:42.539 --> 00:12:45.259 leaps happening in the background. Yes. A week 00:12:45.259 --> 00:12:47.820 before Melissa made landfall, forecasters were 00:12:47.820 --> 00:12:50.720 in agony. The traditional forecast models were 00:12:50.720 --> 00:12:53.159 just disagreeing widely on the storm's eventual 00:12:53.159 --> 00:12:56.259 track. One model had it moving north toward the 00:12:56.259 --> 00:12:58.820 Bahamas, another had it tracking directly into 00:12:58.820 --> 00:13:02.690 the Gulf of Mexico. This uncertainty, that common 00:13:02.690 --> 00:13:06.350 agonizing cone of uncertainty, is a huge challenge 00:13:06.350 --> 00:13:08.570 when you're issuing life -saving warnings. The 00:13:08.570 --> 00:13:10.629 pressure on the human forecasters must have been 00:13:10.629 --> 00:13:13.269 just immense trying to synthesize all those competing 00:13:13.269 --> 00:13:16.070 possibilities. It was. But amidst all that confusion, 00:13:16.649 --> 00:13:19.289 one specific model delivered an outlier forecast 00:13:19.289 --> 00:13:21.610 that proved to be exactly right. One model about 00:13:21.610 --> 00:13:24.029 it. It predicted both Melissa's precise track 00:13:24.029 --> 00:13:27.230 into Jamaica and, crucially, it forecast its 00:13:27.230 --> 00:13:30.559 Category 5 intensity. That model was Google's 00:13:30.559 --> 00:13:33.559 DeepMind AI -based hurricane model. A perfect 00:13:33.559 --> 00:13:36.960 call in a chaotic life or death situation. That 00:13:36.960 --> 00:13:39.200 kind of endorsement for an AI model must have 00:13:39.200 --> 00:13:41.399 been impossible for the meteorological community 00:13:41.399 --> 00:13:44.659 to ignore. It was a watershed moment. Former 00:13:44.659 --> 00:13:46.720 officials from the National Hurricane Center 00:13:46.720 --> 00:13:50.120 later analyzed the 2025 performance of all the 00:13:50.120 --> 00:13:53.019 available guidance. They went on record to call 00:13:53.019 --> 00:13:55.919 the Google DeepMind model the best guidance we 00:13:55.919 --> 00:13:59.490 saw this year. Wow. They noted its stunning accuracy, 00:13:59.750 --> 00:14:01.870 especially when you compare it directly to the 00:14:01.870 --> 00:14:04.230 results from the long -standing traditional physics 00:14:04.230 --> 00:14:07.309 -based systems. That endorsement suggests AI 00:14:07.309 --> 00:14:10.009 isn't just a useful addition anymore. It might 00:14:10.009 --> 00:14:12.370 be fundamentally changing how meteorology even 00:14:12.370 --> 00:14:15.330 works. Let's really unpack the core difference 00:14:15.330 --> 00:14:17.649 for the listener, because these models operate 00:14:17.649 --> 00:14:19.750 on entirely separate philosophies, don't they? 00:14:19.919 --> 00:14:21.620 They really do. You can think of it in terms 00:14:21.620 --> 00:14:23.879 of philosophy. Traditional weather models like 00:14:23.879 --> 00:14:26.279 the standard global forecast system, the GFS 00:14:26.279 --> 00:14:28.620 developed by the National Oceanic and Atmospheric 00:14:28.620 --> 00:14:32.259 Administration, or NOAA, they rely on these complex 00:14:32.259 --> 00:14:35.139 first principle equations. These models are like 00:14:35.139 --> 00:14:38.039 the atmosphere as a physics textbook. They calculate 00:14:38.039 --> 00:14:41.480 how wind, moisture, and heat move and interact 00:14:41.480 --> 00:14:44.179 using established scientific laws. They're trying 00:14:44.179 --> 00:14:46.960 to predict the future based on pure continuous 00:14:46.960 --> 00:14:50.169 mathematical calculation. If an additional physics 00:14:50.169 --> 00:14:53.110 -based model is solving this massive complex 00:14:53.110 --> 00:14:57.269 equation, how does an AI model, like Google DeepMind, 00:14:57.690 --> 00:14:59.950 approach the same problem? It approaches it like 00:14:59.950 --> 00:15:03.029 a detective studying a crime scene archive. AI 00:15:03.029 --> 00:15:05.610 models don't necessarily know physics in that 00:15:05.610 --> 00:15:08.389 traditional sense. Instead, they focus intensely 00:15:08.389 --> 00:15:11.929 on historical patterns. They're trained on decades, 00:15:12.070 --> 00:15:14.850 even centuries of past weather data, looking 00:15:14.850 --> 00:15:17.889 for subtle, nonlinear relationships that a human 00:15:17.889 --> 00:15:20.450 or even a pure physics model might miss. They 00:15:20.450 --> 00:15:23.690 just ask, based on thousands of similar atmospheric 00:15:23.690 --> 00:15:26.029 setups in history, what was the most probable 00:15:26.029 --> 00:15:29.070 outcome? That pattern -based approach explains 00:15:29.070 --> 00:15:32.259 why AI has... for a while now, been pretty good 00:15:32.259 --> 00:15:34.039 at track forecasting. That's essentially a large 00:15:34.039 --> 00:15:35.779 -scale pattern recognition problem. Exactly. 00:15:35.960 --> 00:15:38.320 Storm tracks are governed by these massive, predictable 00:15:38.320 --> 00:15:41.340 atmospheric steering currents. That's why AI 00:15:41.340 --> 00:15:43.879 could always see the path well. But historically, 00:15:44.240 --> 00:15:46.799 AI models always struggled with intensity. And 00:15:46.799 --> 00:15:48.620 why is intensity how strong a storm will be? 00:15:49.000 --> 00:15:50.620 So much harder to predict than where it's going. 00:15:50.840 --> 00:15:53.620 Because intensity is governed by these small, 00:15:53.840 --> 00:15:57.259 chaotic, localized processes. Things like eyewall 00:15:57.259 --> 00:15:59.799 replacement cycles, internal storm dynamics, 00:16:00.159 --> 00:16:02.419 specific wind shear values. Things happening 00:16:02.419 --> 00:16:05.139 on a much smaller scale. Right, and they're notoriously 00:16:05.139 --> 00:16:08.000 difficult to capture in generalized data. But 00:16:08.000 --> 00:16:11.840 the Google DeepMind model overcame this. It incorporated 00:16:11.840 --> 00:16:15.360 highly specific historical data detailing precisely 00:16:15.360 --> 00:16:18.019 how past hurricanes developed and intensified 00:16:18.019 --> 00:16:20.350 under all these various conditions. And that 00:16:20.350 --> 00:16:22.529 allowed it to forecast maximum intensity with 00:16:22.529 --> 00:16:25.250 the accuracy we saw with Melissa. This shift, 00:16:25.610 --> 00:16:27.629 though, while it's delivering better forecasts, 00:16:28.190 --> 00:16:31.129 it introduces a new and kind of challenging dynamic 00:16:31.129 --> 00:16:34.250 for experienced human forecasters. It's this 00:16:34.250 --> 00:16:36.690 critique called the black box problem. It's a 00:16:36.690 --> 00:16:39.289 genuine operational concern. When a forecaster 00:16:39.289 --> 00:16:41.509 uses a physics -based model, they can look at 00:16:41.509 --> 00:16:43.429 the data outputs, the pressure gradients, the 00:16:43.429 --> 00:16:45.610 wind shear fields, and they can trace the precise 00:16:45.610 --> 00:16:47.629 physical mechanism that led to the prediction. 00:16:47.889 --> 00:16:50.330 They know why the storm is being steered northwest. 00:16:50.929 --> 00:16:53.590 But the AI model doesn't offer that same transparency. 00:16:53.789 --> 00:16:57.090 Not always. Because the AI models focus on these 00:16:57.090 --> 00:17:00.090 subtle, learned historical correlations, the 00:17:00.090 --> 00:17:03.809 process can feel opaque. A black box. Massive 00:17:03.809 --> 00:17:06.329 amounts of data go in, a highly accurate forecast 00:17:06.329 --> 00:17:09.769 comes out, but the specific, understandable mechanism 00:17:09.769 --> 00:17:12.470 it used to connect the input to the output might 00:17:12.470 --> 00:17:15.509 be invisible to the human. But if the forecasters 00:17:15.509 --> 00:17:18.250 don't know why the AI made a choice... Let's 00:17:18.250 --> 00:17:20.470 say the AI forecasts a radical path that their 00:17:20.470 --> 00:17:23.009 trusted GFS physics model completely disagrees 00:17:23.009 --> 00:17:25.809 with. How can they trust the AI enough to override 00:17:25.809 --> 00:17:28.799 the physics? Isn't that a massive potentially 00:17:28.799 --> 00:17:32.119 catastrophic liability when human lives are at 00:17:32.119 --> 00:17:34.220 stake. That is the essential challenge of integrating 00:17:34.220 --> 00:17:36.599 AI. It requires a massive leap of faith in the 00:17:36.599 --> 00:17:39.400 underlying statistical accuracy of the AI, even 00:17:39.400 --> 00:17:41.259 without the comfort of that physical traceability. 00:17:41.839 --> 00:17:43.500 And forecasters are inherently skeptical, and 00:17:43.500 --> 00:17:45.180 they should be because they have to be accountable 00:17:45.180 --> 00:17:47.279 for their warnings. The only way you build that 00:17:47.279 --> 00:17:49.380 trust is through repeated, proven performance, 00:17:49.440 --> 00:17:51.839 like the perfect call on Hurricane Melissa. It 00:17:51.839 --> 00:17:54.220 sounds like the human element, the essential 00:17:54.220 --> 00:17:57.220 judgment of the experienced meteorologist, is 00:17:57.220 --> 00:17:59.680 actually becoming even more critical, not less. 00:18:00.099 --> 00:18:02.119 Now they have to decide which computer outlier 00:18:02.119 --> 00:18:05.240 to trust. Absolutely. The consensus is clear. 00:18:05.539 --> 00:18:08.000 AI will not replace the long -standing physics 00:18:08.000 --> 00:18:10.940 -based models or the essential judgment of experienced 00:18:10.940 --> 00:18:14.299 forecasters, but it's clearly going to be a key 00:18:14.299 --> 00:18:16.819 foundational component that provides an alternative 00:18:16.819 --> 00:18:19.359 perspective and critical time savings. It's not 00:18:19.359 --> 00:18:21.539 a replacement. It's a powerful new tool in the 00:18:21.539 --> 00:18:24.279 toolbox. Exactly. And the practical benefit for 00:18:24.279 --> 00:18:27.579 you, the listener, is really simple. As our coastlines 00:18:27.579 --> 00:18:30.000 become more populated, and as these storms become 00:18:30.000 --> 00:18:32.759 more intense, we desperately need more time. 00:18:33.059 --> 00:18:35.319 More time to get people out of the way, to secure 00:18:35.319 --> 00:18:38.619 property, to activate infrastructure. Forecasts 00:18:38.619 --> 00:18:40.859 further and further into the future become vital, 00:18:41.099 --> 00:18:43.259 and AI is delivering that crucial time advantage. 00:18:43.759 --> 00:18:46.299 And this innovation moved beyond the experimental 00:18:46.299 --> 00:18:49.220 phase really fast. We're now transitioning from 00:18:49.220 --> 00:18:52.079 that successful proof of concept stage, like 00:18:52.079 --> 00:18:54.920 Google DeepMind's stellar performance, to large 00:18:54.920 --> 00:18:57.940 -scale operational use deployed by major government 00:18:57.940 --> 00:19:00.960 agencies. Like the National Oceanic and Atmospheric 00:19:00.960 --> 00:19:03.900 Administration. or NOAA. NOAA has now deployed 00:19:03.900 --> 00:19:07.559 a groundbreaking new suite of operational AI 00:19:07.559 --> 00:19:10.539 -driven global weather prediction models as of 00:19:10.539 --> 00:19:14.579 December 2025. This feels like a profound strategic 00:19:14.579 --> 00:19:17.119 leap forward for American weather model innovation. 00:19:17.819 --> 00:19:20.200 It's a signal that the U .S. weather infrastructure 00:19:20.200 --> 00:19:23.299 is fully committing to the AI revolution. It 00:19:23.299 --> 00:19:25.579 really is. The mission statement from NOAA was 00:19:25.579 --> 00:19:28.380 clear and multifaceted. They're aiming to provide 00:19:28.380 --> 00:19:30.980 improved accuracy for large -scale weather and 00:19:30.960 --> 00:19:33.440 tropical tracks, and to deliver that guidance 00:19:33.440 --> 00:19:36.460 much faster and at a dramatically lower cost 00:19:36.460 --> 00:19:38.579 than the current traditional models allow. The 00:19:38.579 --> 00:19:40.859 cost and speed element is the often overlooked 00:19:40.859 --> 00:19:43.079 secret weapon of AI modeling, isn't it? What 00:19:43.079 --> 00:19:45.380 is the most transformative feature of these new 00:19:45.380 --> 00:19:48.200 NOAA AI models compared to the GFS running on 00:19:48.200 --> 00:19:51.019 those massive supercomputers? It is without question 00:19:51.019 --> 00:19:53.279 computational efficiency. We cannot overstate 00:19:53.279 --> 00:19:56.700 this. These AI models use a tiny fraction of 00:19:56.700 --> 00:19:59.079 the computational resources of their traditional 00:19:59.079 --> 00:20:02.089 physics -based counterparts. This isn't just 00:20:02.089 --> 00:20:04.609 about saving money. It's a game changer because 00:20:04.609 --> 00:20:07.430 it allows NOAA to run models more often at higher 00:20:07.430 --> 00:20:09.869 resolution and to run many more members in their 00:20:09.869 --> 00:20:13.750 ensemble systems. It basically democratizes access 00:20:13.750 --> 00:20:16.410 to high quality modeling. Let's look at the specific 00:20:16.410 --> 00:20:18.089 models that NOAA has rolled out. We can start 00:20:18.089 --> 00:20:20.069 with the artificial intelligence global forecast 00:20:20.069 --> 00:20:24.240 system. or AI GFS. This is the single deterministic 00:20:24.240 --> 00:20:27.579 forecast model. Right, so the AI GFS is designed 00:20:27.579 --> 00:20:29.640 to generate a single forecast that's comparable 00:20:29.640 --> 00:20:31.920 in quality to the traditional physics -based 00:20:31.920 --> 00:20:35.119 GFS. And on the performance side, NOAA found 00:20:35.119 --> 00:20:37.819 an immediate gain. improved forecast skill for 00:20:37.819 --> 00:20:40.480 many large -scale atmospheric features, and a 00:20:40.480 --> 00:20:42.900 significant measurable reduction in tropical 00:20:42.900 --> 00:20:45.180 cyclone track errors, especially at longer lead 00:20:45.180 --> 00:20:48.140 times. So just like DeepMind, the AI is delivering 00:20:48.140 --> 00:20:50.200 superior guidance on where the storm will go. 00:20:50.319 --> 00:20:52.460 That's right. Now let's talk about the efficiency, 00:20:52.599 --> 00:20:54.700 because these numbers are just astonishing. They 00:20:54.700 --> 00:20:58.660 redefine what's possible. A single 16 -day forecast 00:20:58.660 --> 00:21:03.559 from the AI GFS uses only 0 .3 % of the computing 00:21:03.559 --> 00:21:06.660 resources required for the traditional GFS. 0 00:21:06.660 --> 00:21:11.019 .3%. Less than half of 1 % of the power. It's 00:21:11.019 --> 00:21:13.119 incredible. And the time savings are just as 00:21:13.119 --> 00:21:15.160 vital. It finishes in approximately 40 minutes. 00:21:15.220 --> 00:21:17.640 40 minutes, compared to a traditional GFS run 00:21:17.640 --> 00:21:20.460 that can take hours and hours. Exactly. That 00:21:20.460 --> 00:21:22.799 dramatically reduced latency means forecasters 00:21:22.799 --> 00:21:25.980 get critical data much, much faster. Speed means 00:21:25.980 --> 00:21:28.500 earlier warnings, and earlier warnings mean lives 00:21:28.500 --> 00:21:32.079 saved. That speed advantage is huge. But we did 00:21:32.079 --> 00:21:34.400 note there was one initial area for improvement 00:21:34.400 --> 00:21:38.420 in this first version, V1 .0. Yes. The initial 00:21:38.420 --> 00:21:40.640 testing confirmed that prevailing challenge in 00:21:40.640 --> 00:21:43.660 AI. Version 1 .0 did show a slight degradation 00:21:43.660 --> 00:21:46.200 in tropical cyclone intensity forecasts compared 00:21:46.200 --> 00:21:48.460 to the traditional physics -based model. Still 00:21:48.460 --> 00:21:51.240 the hardest nut to crack. It is. But the speed 00:21:51.240 --> 00:21:53.619 gains and the track improvements were so significant 00:21:53.619 --> 00:21:55.759 that they deployed it operationally immediately, 00:21:56.099 --> 00:21:57.779 with the understanding that future versions will 00:21:57.779 --> 00:22:00.279 focus heavily on boosting that intensity accuracy. 00:22:00.640 --> 00:22:03.759 Okay, next up is the really essential component 00:22:03.759 --> 00:22:07.900 of modern forecasting. The ensemble system. NOAA 00:22:07.900 --> 00:22:11.000 deployed the AIG -FS, the artificial intelligence 00:22:11.000 --> 00:22:13.880 global ensemble forecast system. Before we get 00:22:13.880 --> 00:22:16.240 into the AI version, just remind us, what is 00:22:16.240 --> 00:22:19.309 the core role of an ensemble system? An ensemble 00:22:19.309 --> 00:22:21.950 system is maybe the most crucial tool forecasters 00:22:21.950 --> 00:22:24.470 have for managing uncertainty. Instead of running 00:22:24.470 --> 00:22:27.210 the model just once, you run it many times. In 00:22:27.210 --> 00:22:30.529 Noah's case, 31 times. And each run starts with 00:22:30.529 --> 00:22:33.009 slightly different initial atmospheric conditions. 00:22:33.049 --> 00:22:34.589 So you get a range of possibilities. You get 00:22:34.589 --> 00:22:37.150 a fan of possibilities, a range of probable outcomes, 00:22:37.269 --> 00:22:40.450 rather than one single confident solution. It 00:22:40.450 --> 00:22:42.750 helps meteorologists and decision makers understand 00:22:42.750 --> 00:22:45.069 the overall forecast uncertainty and figure out 00:22:45.069 --> 00:22:47.890 the highest risk scenarios. And how did the AI 00:22:47.920 --> 00:22:51.099 based AGFS perform against the traditional global 00:22:51.099 --> 00:22:55.079 ensemble forecast system, the GEFS. The AI version's 00:22:55.079 --> 00:22:56.920 forecast skill is comparable to the traditional 00:22:56.920 --> 00:22:59.000 ensemble system, so the quality is maintained, 00:22:59.079 --> 00:23:01.920 but it provides a huge benefit. It extends reliable 00:23:01.920 --> 00:23:04.440 forecast skill by an additional 18 to 24 hours. 00:23:04.460 --> 00:23:07.180 A full day. An enormous gain in lead time. And 00:23:07.180 --> 00:23:09.759 again, the efficiency is astounding. It requires 00:23:09.759 --> 00:23:13.059 only 9 % of the computing resources of the traditional 00:23:13.059 --> 00:23:16.299 GEFS. Those savings and time gains are phenomenal. 00:23:16.819 --> 00:23:19.759 But the real innovation breakthrough, the thing 00:23:19.759 --> 00:23:22.099 that makes NOAA's deployment a global first, 00:23:22.799 --> 00:23:27.460 is the HGEFS, the hybrid GEFS. This sounds like 00:23:27.460 --> 00:23:29.299 the ultimate answer to that black box problem 00:23:29.299 --> 00:23:31.400 we were talking about. This is where the physics 00:23:31.400 --> 00:23:33.759 and the patterns marry. It acknowledges that 00:23:33.759 --> 00:23:37.059 neither system is perfect on its own. The HGEFS 00:23:37.059 --> 00:23:40.539 is a pioneering 62 -member grand ensemble. Okay, 00:23:40.559 --> 00:23:42.380 what does that mean? It's created by combining 00:23:42.380 --> 00:23:45.380 the 31 members of the physical GEFS ensemble 00:23:45.380 --> 00:23:48.740 with the 31 members of the AI -based AIGF. So 00:23:48.740 --> 00:23:50.779 the hybrid model is essentially letting the physics 00:23:50.779 --> 00:23:53.200 tell us the rules of the atmosphere. while the 00:23:53.200 --> 00:23:55.720 AI uses the patterns of history to refine those 00:23:55.720 --> 00:23:59.339 rules. All in one gigantic, robust package. That's 00:23:59.339 --> 00:24:01.680 a perfect way to describe it. By combining two 00:24:01.680 --> 00:24:04.200 fundamentally different modeling systems, one 00:24:04.200 --> 00:24:06.880 based purely on physical equations, one based 00:24:06.880 --> 00:24:10.940 on learned historical patterns, the HGEFS creates 00:24:10.940 --> 00:24:13.279 an ensemble that's twice the size, making it 00:24:13.279 --> 00:24:15.920 much more robust. And crucially, because the 00:24:15.920 --> 00:24:18.740 two systems fail for different reasons, combining 00:24:18.740 --> 00:24:21.019 them more effectively represents the full spectrum 00:24:21.019 --> 00:24:23.519 of forecast. uncertainty. And the result of that 00:24:23.519 --> 00:24:26.319 blending, did it actually work to create better 00:24:26.319 --> 00:24:29.720 forecasts? It consistently worked. The HGEFS 00:24:29.720 --> 00:24:32.660 consistently outperforms both the AI -only system 00:24:32.660 --> 00:24:35.440 and the physics -only system across most major 00:24:35.440 --> 00:24:38.039 verification metrics. It is a first -of -its 00:24:38.039 --> 00:24:40.960 -kind operational approach globally, and it confirms 00:24:40.960 --> 00:24:43.559 that the future of forecasting is hybrid. It's 00:24:43.559 --> 00:24:45.880 physics -informed by history. All run with maximum 00:24:45.880 --> 00:24:48.279 computational efficiency. And this entire suite, 00:24:48.559 --> 00:24:52.619 the AIGFS, AGSS and HGEFS. This wasn't just whipped 00:24:52.619 --> 00:24:54.980 up overnight. This is the culmination of a huge 00:24:54.980 --> 00:24:57.119 collaborative effort. A massive undertaking. 00:24:57.380 --> 00:24:59.859 It resulted from something called Project EGL, 00:25:00.519 --> 00:25:02.500 which was a collaborative initiative involving 00:25:02.500 --> 00:25:04.660 various elements of NOAA, the National Weather 00:25:04.660 --> 00:25:07.359 Service, its Oceanic and Atmospheric Research 00:25:07.359 --> 00:25:09.740 Labs, the Environmental Modeling Center, and 00:25:09.740 --> 00:25:12.200 the Earth Prediction Innovation Center. So they 00:25:12.200 --> 00:25:14.519 leveraged work from the private sector too. They 00:25:14.519 --> 00:25:17.140 leveraged foundational academic and private industry 00:25:17.140 --> 00:25:20.160 work, specifically Google DeepMind's Breakthrough 00:25:20.160 --> 00:25:22.680 Graphcast model, and then they fine -tuned it 00:25:22.680 --> 00:25:25.859 using NOAA's own massive global data assimilation 00:25:25.859 --> 00:25:28.980 system data. So they took the best of theoretical 00:25:28.980 --> 00:25:31.940 private industry innovation and married it with 00:25:31.940 --> 00:25:34.759 the essential operational necessities and the 00:25:34.759 --> 00:25:38.240 vast real -time data flow of the National Oceanic 00:25:38.240 --> 00:25:41.299 and Atmospheric Administration. That is how you 00:25:41.299 --> 00:25:43.880 achieve a quantum leap in forecasting capability. 00:25:44.160 --> 00:25:46.099 The administrator of NOAA summed it up perfectly. 00:25:46.420 --> 00:25:48.819 This is a new paradigm for providing improved 00:25:48.819 --> 00:25:51.539 accuracy, faster delivery of forecast products, 00:25:51.599 --> 00:25:53.920 and at a dramatically lower operational cost. 00:25:54.680 --> 00:25:56.920 The pace of innovation is finally accelerating 00:25:56.920 --> 00:25:59.619 to meet the pace of climate change. As we wrap 00:25:59.619 --> 00:26:01.980 up this deep dive into 2025, let's work all back 00:26:01.980 --> 00:26:05.039 together into a single narrative. From the persistent, 00:26:05.579 --> 00:26:07.880 overwhelming heat, despite the natural cooling 00:26:07.880 --> 00:26:10.599 efforts of La Nina, which showed us the relentless 00:26:10.599 --> 00:26:14.150 nature of human driven warming. To the profound 00:26:14.150 --> 00:26:16.269 social inequities highlighted in regions like 00:26:16.269 --> 00:26:20.109 South Sudan, 2025 really showed us the current 00:26:20.109 --> 00:26:22.829 harsh reality of a planet that is fundamentally 00:26:22.829 --> 00:26:26.069 warmer and more volatile. And the devastating 00:26:26.069 --> 00:26:28.849 power of Hurricane Melissa in the Gulf of Mexico 00:26:28.849 --> 00:26:31.450 region was the perfect, terrifying case study 00:26:31.450 --> 00:26:34.190 of that. It reminded us that the ultimate devastation 00:26:34.190 --> 00:26:36.849 is still possible and that even the best preparation 00:26:36.849 --> 00:26:39.390 has its limits when a storm achieves that kind 00:26:39.390 --> 00:26:42.750 of maximum intensity. But the rapid pace of innovation, 00:26:43.029 --> 00:26:45.130 evidenced by both the success of Google DeepMind 00:26:45.130 --> 00:26:47.269 and the operational deployment of NOAA's hybrid 00:26:47.269 --> 00:26:49.430 AI models, it suggests that while the weather 00:26:49.430 --> 00:26:52.309 is becoming undeniably more extreme, our ability 00:26:52.309 --> 00:26:55.190 to see it coming and faster is reaching astonishing 00:26:55.190 --> 00:26:58.089 new heights. It really is. It's offering that 00:26:58.089 --> 00:27:02.569 extra vital 18 to 24 hours of warning time. It's 00:27:02.569 --> 00:27:05.190 a race against the clock. As the climate changes 00:27:05.190 --> 00:27:08.029 and volatility increases, our tools for prediction 00:27:08.029 --> 00:27:11.170 have to evolve even faster just to keep pace. 00:27:11.990 --> 00:27:14.009 I want to leave you with this final provocative 00:27:14.009 --> 00:27:17.650 thought. While forecast accuracy is dramatically 00:27:17.650 --> 00:27:20.009 improving and we celebrate every single hour 00:27:20.009 --> 00:27:22.349 of extra lead time, Hurricane Melissa reminded 00:27:22.349 --> 00:27:25.190 us that adaptation, no matter how good, has hard 00:27:25.190 --> 00:27:28.069 limits. Knowing a category 5 is coming earlier 00:27:28.069 --> 00:27:30.710 is absolutely vital for saving lives, but what 00:27:30.710 --> 00:27:33.690 truly prevents catastrophe? Is it only forecasting? 00:27:34.309 --> 00:27:36.549 Or is it fundamentally reducing the fuel for 00:27:36.549 --> 00:27:38.750 these storms, addressing the root cause of the 00:27:38.750 --> 00:27:40.450 warming that makes them so powerful in the first 00:27:40.450 --> 00:27:42.150 place? That's the question we have to grapple 00:27:42.150 --> 00:27:44.150 with moving forward. If you want to keep exploring 00:27:44.150 --> 00:27:46.130 the science, chaos, and stories behind the weather, 00:27:46.349 --> 00:27:48.849 please like, follow, comment, and rate the Meteorology 00:27:48.849 --> 00:27:51.549 Matters podcast. You can also follow meteorologist 00:27:51.549 --> 00:27:54.710 Rob Jones on Instagram at Meteorologist, on TikTok 00:27:54.710 --> 00:27:57.430 at TVMeteorologist, and on YouTube by searching 00:27:57.430 --> 00:27:59.670 for Rob Jones Hurricane, where you can also find 00:27:59.670 --> 00:28:02.200 the Meteorology Matters podcast playlist. You've 00:28:02.200 --> 00:28:04.799 been listening to Meteorology Matters, created 00:28:04.799 --> 00:28:07.539 by meteorologist Rob Jones. Thanks for listening, 00:28:07.900 --> 00:28:10.380 and remember, meteorology always matters.