Moving on from Hanna, as my focus is on forecasting. To be clear, Hanna remains a major issue for the Eastern U.S. and Southeastern Canada, with respect to heavy surf, beach flooding, heavy rain, and gusty winds. But the forecasting perspective is essentially done with. So, we move on to Ike.
Ah, crow, a forecasters tastiest meal. Two days ago I said that it would be very unlikely for Ike to push west of the Florida Panhandle... and even as far west as the Panhandle looked a bit unlikely. I even tussled with a commenter about that. And, frankly, under the same circumstances, I'd make the same forecast. Why? Because it is uncommon to see all long range models blow a forecast in the same direction. Typically, long range errors from forecasters is because we pick the wrong model or model cluster to trust. This time, all models from a few days ago blew this - or so it seems (given the big change, we need to be on alert for another shift). We'll address why the models would shift like this, and get to the forecast after the jump...
So, how could 100% of the models agree... all showing a turn north in the extreme Eastern Gulf, over Florida, or offshore of the U.S., with little or no threat whatsoever to Cuba and Hispaniola, as we saw a few days ago... and 100% of them be wrong? Granted, that was a pretty long-range forecast and errors are to be expected. But, typically, there is a wide model spread and at least a few models usually have the right idea. Or, if all are wrong, they're not usually all wrong in the same direction like this. What can cause this? A typical cause of this is issues with data input. Erroneous, but "not so bad" data is the bane of weather models. It is impossible (well, perhaps a close post-analysis could find the problem ...but at this stage it's "impossible") to say for sure if bad data input was, indeed, the problem, and what data it was. But, I suspect marginally poor data from a rather data sparse region got into the models many days ago. It would have to be only marginally poor, or the models' quality control preprocessors would've thrown it out. And it would have to have been in a data sparse region, because the bad forecasts continued repeatedly... if lots of other good data was around, this bad data would've been quickly superceded (note that model initializations fill in data gaps with the previous model run, so in data sparse regions, the model tends to look the same run after run - just projecting things forward). Of course, everything moves along in the atmosphere. So, eventually, this "parcel" of air that got poorly analyzed due to the bad data will get into a region where there are radiosondes or other high quality observations. At that point, "good" data will get into the models and correct the error, fixing the model solutions.
You may then ask how the European model was the first to get "fixed"? Well, that could be any number of reasons. Some model initializations are more "stubborn" and could take several runs of the "good" data before they correct. Also, the European model, as we've discussed, has a westward bias... and since that happens to be the same direction as the correction, it may have just gotten lucky. But it could have legitimately picked up on the change sooner, because the European weather service does a better job of incorporating satellite data into their models - which obviously would give them better coverage of quality data (though satellite data is nowhere near the quality of radiosondes... but if used properly, the relatively high resolution of that data can certainly nudge the European model in the right direction).
Okay, that all said, what does it mean for the forecasting going forward for Ike. Well, first let me emphasize that there's still a LOT of uncertainty, and I may be able to crow on ice (still some model guidance - even though south and west early in the period - still turns Ike north with the same cold front as previously expected, and into western Florida... so, technically, not really much change in the existing forecast, despite all my bellyaching... just a much greater Cuba threat and a much reduced south Florida threat). But there is some significant model spread, all the way from an eventual landfall near Brownsville, TX around through Florida. But that is a long way off, with huge question marks, and continued west trending on some models makes one wonder whether or not the landfall will shift more on some models - down into Mexico. The near term forecast, however, is in better focus, and is also shifting south.
This more southward shift is supported by ongoing trends. Latest satellite loops continue to show Ike losing latitude (i.e., shifting southward) as he moves west. Here's how he looks right now:
...you can see in both the loop and that image that Ike will, unfortunately, probably brush by Hispaniola close enough to bring some outer bands with heavy rainfall to Haiti. If you've been following this and other diaries on dKos, you know that Haiti was hit hard by Hanna, with a significant death toll and slow recovery. The last thing they need is Ike. Likely, they won't get a direct, head-on shot from him. But, they will likely get some heavy rain - especially in northwestern Haiti, the same region hammered by Hanna.
Thereafter, we begin to get some spread in the guidance. Here's what most of the models looks like (though some are not on this graphic and others have been updated since this graphic was generated):
...and I should point out the "missing" lines on that plot would actually show increased spread among the models. For example, the European and Canadian models, not on this plot, support the southernmost solutions - as seen by the EGRI (UK model) solution on that map... with eventual landfalls (beyond the time period of the map) near Brownsville, TX. But on the flip side, the newer runs of the GFDL (GFDI on that map), HWRF (HWFI), GFS (AVNI), and NOGAPS (NGPI) have all shifted east. That plot shows the overnight model runs, while for those four models, there has been a more recent early morning run. The early morning GFDL aims Ike towards north of Tampa (landfall not indicated by the end of the forecast... but getting close), on Thursday. The new HWRF is slower, so impossible to project its landfall, but likely - given the position and motion at the end of its forecast - somewhere on the west coast of Florida. The new GFS has some gyrations in the east or central Gulf, but eventually curls Ike northeastward to the Big Bend of Florida by late next Saturday. And the new NOGAPS model is similar to the GFS, but with no dawdling around, has its landfall Friday morning.
So, we've got a huge range of model solutions. And, frankly, I'd have to admit, I have no clue which will verify. First, one thing is clear, Ike will come close enough to Cuba, and probably make a direct hit, at least on the north coast. So, Cuba will get slammed late tomorrow and into Monday. Beyond that, I'm honestly not sure. Here's what the National Hurricane Center shows:
...they're obviously splitting the difference between the two groups of models. We have two "camps"... the southern Euro, Canadian, and UK model... and the northern GFS, NOGAPS, HWRF and GFDL. NHC is taking the safe bet approach. Can't say that I blame them. It's probably the smartest approach. But, in reality usually it is either one group or the other that is correct. So, don't pay strict attention to the trajectory of that above map at the end of the day five period. It shows a direction aiming, generally, towards the northwestern Gulf. But please be aware that all locales from Brownsville, or even northern Mexico, all the way around to Tampa are under an equal risk. And I can't really tell you which model "camp" will prevail. My thoughts on the matter are ambivalent. On the one hand, current motion, model trends, and model "success" (relatively ...and apparently ...by the Euro) strongly favors the southern track. On the other hand, the early morning model runs have all reversed the trend, the Euro model has a westward bias, and the Euro model is the only "quality" model (based on this season's performance) in the southern group. This all strongly favors the more north track... At this point, I'm not even going to try to guess. All interests in the Gulf - east, central and west - stay on alert. The only seemingly sure bet is Cuba getting hit hard (landfall or not... though landfall seems more likely than not) and south Florida seeing at least some fringe effects - possibly significant effects in the Keys.
Intensity also gets severely complicated by landfall, etc. Current recon data and the above shown satellite imagery indicates that shear took a toll on Ike, but not as severely as expected, and should be abating as we speak. So, Ike remains a Category 3 hurricane, and re-intensification could begin at any time (in fact, the most recent recon data hints that that could be occurring already... but that is not a sure thing, as an official "fix" has not been made yet). But interaction with Cuba is a big, big question mark with respect to intensity. The intensity models show the following:
...if and how long Ike moves over Cuba will be a huge factor in his intensity beyond that point. However, giving some confidence in the longer range forecast is the fact that the new HWRF and GFDL (not on that plot), which both have landfall on Cuba, both shows conditions in the Gulf good enough for a rebound to Category 3, but then stalling beyond that. Obviously, it's not "good" for anyone to get a Category 3 hurricane. But the "good" thing from a forecasting perspective it that this seems to tie Ike into about a Category 3 once he's in the Gulf. Intensity forecasts are terrible, so I wouldn't nail that down. But the point is, models and an analysis of upper level conditions clearly support Ike rebounding off of Cuba, but conditions also seem to imply that he'll be capped. So, we can have fair confidence that Ike will be neither a minimal hurricane nor a Category 5 after his post-Cuba recovery... by mid-week... once he's into the Gulf.
Getting this done late. So, I'll end it there and get this posted. The bottom line is that the southern Bahamas and Cuba are under the most certain, direct, severe threat over the next few days. And Haiti, though they'll miss the worst of it, will see rain in the least-desired area... the northwest part of the island. Beyond that, Ike goes into the Gulf, but thereafter, it's anoyone's guess where Ike goes... but it's nearly impossible to get out of the Gulf without hitting somewhere. So, barring dissipation (highly unlikely) Gulf residents, stay alert.
UPDATE 3:30PM EDT My apologies for the late, boring (graphics-free) update. But I'm rather "out of pocket" (I hate that saying, who came up with that?) today. But I must get this update out as there is some important new data in for Hurricane Ike...
First, let me address his current/recent intensity. Ike has been very, very slowly intensifying all day. I rarely take NHC to task... they're in an unenviable position having such huge responsibility under such tight budget, time and personnel constraints. But they blew this by not waiting for the aircraft reconnaissance data. Based on what, I'm not sure... in their 11AM EDT advisory, they lowered Ike to 95kts. The recon fix came in six minutes after advisory time. And they knew it was coming (they're scheduled). They should have waited (they have, in the past, frequently run a few minutes late with advisories - no one would have gotten bent out of shape if they waited a few minutes for the new data, then quickly tweaked their forecast back into shape). The recon data came in and totally blew their call. The recon fix hit winds of 125kts (145mph) at flight level. That supports winds of 105+kts (120-125mph) at the surface. And their onboard instrumentation (SFMR ...Stepped Frequency Microwave Radiometer) estimated winds just over 100kts. The SMFR, though a complex peice of equipment, has proven itself highly imperfect. So, 105kts would've been the best setting for Ike. But, fine, even leaving Ike at 100kts would've been fine. Instead, NHC dropped him to 95kts. He was never 95kts. He was holding steady at 100kts, or increases slightly to 105kts.
Since then, the next recon fix got a pressure 3mb lower, flight level winds 2kts higher, and SFMR winds 3kts higher. So, Ike has clearly been slowly gaining strength. This data, even the SFMR, all supports Ike being at least a 105kt hurricane, possibly even 110kts. And, now, in the National Hurricane Center's intermediate 2PM advisory on Ike, they've taken winds back up to 100kts (100mph). So, now, they're running 5-10kts behind.
To be fair, I'm splitting hairs a bit here... 5kts is pretty meaningless. But the problem is, the media spin on the 95kts was ugly. It indicated weakening and took Ike below "major hurricane" Category 3 status. Though I'm "out of pocket", I did have access to some media, and they were ballyhooing the downgrade. Simply put, the downgrade was superficial and erroneous. Ike has not weakened. He has strengthened. And I am not saying this to be alarmist - in fact, while this is nothing but bad news for Cuba, this changes nothing in the thinking for U.S. interests that Ike will get "capped" at a Category 3 in the Gulf (though I'd also continue to warn strongly about how poor intensity forecast is... it is my best guess that he'll be capped at a Cat 3, but I'll be the last one shocked if he hits a four, especially when he crosses the loop current).
The other big news is the afternoon model runs. Modest changes in almost all of the models and, importantly, all towards one another!! We have much, much, much better model agreement this afternoon. I would, once again, warn strongly that U.S. landfall concerns remain 6+ days away and, so, forecast errors are typically large. And model agreement does not mean model correctness... recall, they all agreed a few days ago on a track to eastern Florida then up the East Coast or out to sea. Yeah, that panned out well < /sarcasm>.
No good graphic summation of the afternoon models is available yet. So, here's just a quick, boring rundown of what each shows:
UKMET: Slight northward shift towards south Texas on Friday (this solution is now an outlier).
Euro: Big north shift to near landfall in Port Arthur, TX next Saturday - though it doesn't actually show landfall, spinning Ike down the Texas coast thereafter... but the early shift north is critical.
GFS/American: Landfall near Pensacola next Saturday.
Canadian: Approaching landfall around central or eastern Louisiana next Friday or Saturday.
Navy: Into the north-central Gulf next weekend, but drifting - without landfall.
GFDL: No landfall by day five, but aiming towards somewhere between New Orleans and Pensacola for, roughly, day 6 or 7.
HWRF: Slower, tough to make any conclusions based on its final, day 5, position.
Bottom line: Model agreement is near unanimous, save one outlier, on getting Ike into the northern Gulf coast in the Friday/Saturday timeframe. There are now zero solutions east of Pensacola, and only 1.5 solutions (UKMET + 0.5 for the Euro's late-period swing down the coast) west/south of Beaumont/Port Arthur, TX. That's a pretty tight range for a day six or seven forecast. Confidence in this is definitely not high! All residents, everywhere along the Gulf Coast should remain on alert!! But there is at least now some general idea that the threat may be greatest along the north-central and north-west Gulf Coast.
The lone sure-bet now seems to continue to be that Cuba is going to get hit hard by Ike. And, honestly, there is nothing good to hope for here. Ike may move right along the spine of Cuba. But there's a chance it'll bump offshore a bit faster. Either has its bad effects. For the sake of Cuba we may hope that it'll come offshore faster and pull away, lessening the impact on the island. But that will allow Ike to get stronger, faster - a potential problem for the U.S. On the flip side, if Ike runs the entire length of Cuba, he may not be able to recover, making him a much lesser threat to the U.S.... but that'll create a miserable situation for Cuba.
...and, of course, that Cuba question is highly relevant for the issue of how strong Ike may be when he makes his final landfall (likely in the U.S., but I'll assume nothing with this storm). I'm sticking to the thinking in the original post - that he'll get capped around a three. So, I'm anticipating that wherever the final landfall is, Ike will be on the weak end of the Category 3 range. With, one last time, a strong, strong emphasis on the fact that intensity forecasting is poor, especially that far out. So, that's merely an educated guess.
Final note before "going to press"... new recon fix in... slow, slow intensification continues. Pressure down further (950mb). Flight level winds up further (129kts). SFMR was lower this time (99kts). But the pressure is the best indicator. So, intensification is ongoing, albeit very slow. And, based on the flight level winds, Ike is now at 110kts... though I wouldn't be at all surprised if NHC ups him only to 105kts at 5PM EDT... given that SFMR has only peaked at 104kts. Of course, I'm no mind reader, so I could be wrong. Plus, more recon data will come in prior to their next update - so, some new, important info could push them one way or the other.
That's it for now. Sorry for the "dull read" in this update.