: : d u r a : m a t e r :

Wednesday, December 29, 2004

Banda Aceh:

Today's Sumatra -Andaman Islands event was one of the great earthquakes - we will see only one or two like this in our lifetimes. The combination of the location and orientation made for a horrendous result - several tens of thousands dead, with a rising death toll guaranteed. (This post took a while to finish - so it contains material from well after the event...)

First, the tectonic setting (click on the image for larger version):

The India and Australia plates are moving northward, and are shoving under (subducting) the Burma microplate. Here's a schematic of the setting, with the India/Australia plate on the left, moving to the right and under the Burma microplate on the right:

The movement occurs in sudden bursts -- earthquakes. Like banging a tray full of water, if the earthquake moves the seafloor, it sets off waves -- tsunamis.

Here is one of the early simulations done of the tsunami by the Japan National Institute of Advanced Industrial Science and Technology:

And here is another simulation of the Sumatra-Andaman tsunami done by NOAA. Notice the reflections off and refractions around India, Diego Garcia, the 90 East Ridge and the Maldives! These reflections and refractions can focus energy and cause the tsunami to be very high on some shores and not so high on others - so the threat is not uniform along all shorelines. Note that Australia's West coast is much closer to the initial movement than Kenya, but it got almost no resulting wave.

The Geological Survey and Caltech's Seismo Lab produced a first estimate of the amount of the slip along the fault of the great Sumatra-Andaman Islands earthquake:

The red star is the location of the hypocentre - where the earthquake began. The black line is the surface location of the subduction fault (the trench), colours show the amount of displacement in cm of the hanging (upper) wall of the fault relative to the foot (lower) wall. Note that most maps show the hypocentre as if this was where all the action was, but that a lot of the motion of the seafloor was probably several hundred kilometers to the Northeast (and in fact, later modelling shows that there was significant motion even up into the Andaman and Nicobar Islands). This difference between the location of the hypocenter and the main displacement values is typical - and misleading.

Now it becomes obvious why the North-West of Aceh got hit so hard - the largest slip is 200km farther up the peninsula than the hypocentre.

Note that even in this preliminary model, an area of about 1200 km by 100km has moved several meters. This resulted in several meters of seafloor uplift (and displaced water, creating the wave). The total energy released was estimated at 3.57X10^29 dyn.cm in 200 seconds, with a peak slip of about 20 meters. (Chen Ji, Caltech).

There were at least fourteen aftershocks larger than magnitude 6.0, and one larger than 7.0 (all of these respectable 'quakes on their own).

This was enough to cause a 22cm wave in San Diego harbor, 2.6 meters in Manzanillo, Mexico due to focusing effects, and several days of tidal height disturbances as the sloshing continues to reverberate around the ocean basins.

Movie of the Andreanof tsunami movie
- moves much more slowly in the shallow waters north of the Aleutians
- reflects off/refracts through the Hawaiian/Emperor chain - the BC coast gets hit twice: once directly, the second time as a reflection off the Hawaiian seamounts.

What it looked like on the ground in Thailand (Aftonbladet, 8.0 Mb .mov file)This is a different angle from a clip that received a lot of air time. During the second surge, off to the lower left while looking out to sea, one can see a couple behind a palm tree trying to get out from behind a pile of debris blocking the surge and into the bar area. As with the shot from the other angle, I'm not sure they made it before the debris collapsed onto them, and pinned them under tons of rushing water. This is a horrendous example of how innocent looking water can quickly turn deadly. This begins as an interesting video, and quickly becomes quite frightening.

Links:

There is a lot of talk now about an Indian Ocean Tsunami warning system, and support from the U.S. Government for a global system that will include the Atlantic too. Perhaps a good idea, but several things should be noted. The frequencies of tsunamis in these two other oceans are very much less than in the Pacific - so there is a very high probability that the systems will never be used. On the other hand, the process of preparation of such a system reduces societies' vulnerability to many different types of hazards at the same time, so while we may not use them for saving people from tsunamis, we may end up using them to save people from other types of disasters. The last thing is a comment regarding tsunamis - even if there had been a fully operating system in place (which includes not only an earthquake detection system and a wave detection system, but more critically an information dissemination and public evacuation plan), the system would have done nothing to save Aceh or possibly even Thailand, since the wave would have reached them before any possible plan could have been activated.

Here is an amazingly prescient paper published in the September 2004 issue of Geoscience Australia magazine about an Indian Ocean tsunami warning system: AusGeo News 75 p.4 (part 1, with the referenced article, is a 3.9 Mb PDF file)

The best advice: if you see the sea withdraw, you can be guaranteed that it will be coming back with a vengeance. Run for the hills.

And finally, a word about feral goats in the Andamans, who, I would bet, are still there.

Wednesday, December 22, 2004

Oh no, a physics post! It even has... formulas!

Our day-to-day experience with weights and speeds is limited to a small range. We don't expect the behaviour of objects to change very much, even when the object is far outside this limited range of weight or speed.

We know very well that a small piece of lead (between 2 and 30 grams), travelling very fast (between 300 and 1,000 meters per second), can do a great deal of damage - i.e. a bullet.

However, when we think of a large piece of polystyrene foam (about 30 cm across), then all bets are off. After all, it's hard to think of being able to hurl a large piece of foam fast enough to do damage to anything, right?

I went to a lecture the other day by Doug Osheroff that demonstrated this very clearly, using some high-school level physics.

Doug has a Nobel Prize. He's a low-temperature physicist who works at Stanford, and who recently served on NASA's Columbia Accident Investigation Board. So I paid attention to what he was saying.

He outlined the basic calculations behind the conclusion that a 0.75 kg chunk of foam from the Shuttle's external tank (ET) had irreparably damaged the leading edge of Columbia's left wing.

Video of the launch taken from various angles shows the foam breaking off and striking the wing - this allows an estimation of the source of the foam (the -Y bipod ramp, and therefore its size & weight, +/- 30cm radius, 0.75 kg), and also the amount of time it took to travel the distance from the ramp to the leading edge (19 meters in about 0.16 seconds).

The key here is to realize that since the foam fell off the external tank, it was decelerating relative to the air rushing by at Mach 2.5, and accelerating relative to the shuttle (and its onrushing wing).

So:
d₀ = ½at₀²

Where d₀ is the distance travelled, a is the acceleration of the foam, and t₀ is the time between the foam breaking off and the wing strike.

Substituting the values:

19 m = ½ a (0.16 sec)²

gives

a = 1,480 m/s²

or about 150 g's.

How fast was that foam travelling when it got to the wing?

v₀ = a t₀

substituting values:

v₀ = 1,480 m/s² · 0.16 sec

gives

v₀ = 238 m/s

Converting to something we understand better: 855 km/h or 530 mph. Pretty fast.

But it's still hard to comprehend a chunk of foam doing such tremendous damage.

In terms of kinetic energy,

E_k = ½ m v₀²

plugging in our values:

E_k = ½ 0.75 kg· (238 m/s)²

gives

E_k = 21,240 N·m (or J)

Say we had a bullet with equivalent kinetic energy - how fast would it have to be going? Let's choose the a fairly standard .22 bullet of about 40 grains weight, or about .0026 kg

Then:

21,240 J = ½ 0.0026 kg · v₁²

gives

v₁ = 4,040 m/s

or about 13,260 fps, which is about five times faster than a bullet coming out of a rifle - or about twenty-five times the energy a regular 0.22 bullet would pack.

OK, obviously were talking something serious here.

And what was it crashing against? Aren't those wings tough? After all, they can take flying through plasma, right?

Umm, well, no. In fact, the leading edge material is really fragile. The RCC panels on the shuttle have to be covered with protective padding when the shuttle is being serviced to avoid damage from dropped tools or even dropped bolts.

The plasma-resistant wing never stood a chance against a chunk of foam.

Monday, November 29, 2004

Farewell to the NeXT world:

Steve Jobs has cut a swath through my life. Each time I have bought hardware, he has cancelled its production within a few months (or 'Steved' it, as it is known in the Mac/NeXT/Newton communities).

I bought my NeXT colorstation slab on April 3, 1992. NeXT ("black") hardware was Steved in October of 1993.

I bought my Apple Newton MP2100 in February of 1998. Newton ("green") hardware was Steved the same month.

I'm not sure whether to classify myself as an 'early adopter' or an idiot. Either way, the products still perform well. The NeXT, while slow, can handle most any attachment or compression scheme that my Macs or PCs cannot. I still use several Newtons.

...sigh. Now I'm just waiting for the axe to fall on my latest purchase, the iMac. I once thought of posting news of what I had recently bought so that everyone could stay away from it. At least I am confident that my other purchase, the iPod, is here to stay.

Why did I buy the NeXT in the first place? Well, I had just finished my Ph.D. at MIT, and I had a ton of UNIX tapes with my data and codes, a thesis written on a Mac, and I could see Microsoft's tendrils reaching out and choking out the future. So when I saw this cold, cool, black machine running Mac, DOS, and UNIX windows at the same time on a 4,096 color screen (and you could Cut & Paste between them!) I jumped. Here's my recreation of what I saw that day (remember, this is 1992...):

Sold!

This ease of use and versatility is why Tim Berners Lee decided to develop what would become the world-wide web on a NeXT while he was at CERN.

In the process of finally packing up my NeXT machine, I came across some interesting memories:

Remember Mac 800 kb diskettes? Well, the NeXT came with a drive capable of reading 2.8 Mb Extended Density (ED) diskettes. I've got dozens of boxes of the 800 kb diskettes, and a couple of boxes-worth of the 2.8 Mb ones. And no way to read any of them, since none of my remaining floppy drives will handle anything but the 1.4 Mb standard Double Density format.

2.8 Mb ED diskette

As I mentioned in my previous NeXT-lamentation post, I have enjoyed seeing traces of the NeXT in Mac OS X. Before I packed up the NeXT machine, I pulled off as many icons from the system as I could.

Here are four icons from the NeXT and Mac OSX that give away the heritage:

NeXT and OSX Home Folder icons

This next one is a little more difficult to catch, since they are for different applications:

NeXT Prefs and OSX digital clock

Jobs left Apple in the late 1980's to form NeXT, and it has been said that he wanted to crush Apple with a vastly superior product. He did create a vastly superior product, but the price was so high that he could not get NeXT unit sales to a sufficient level to satisfy creditors, and in a move oddly similar to last week's IBM sell-off of its PC manufacturing facilities to the Chinese, NeXT ended up selling its hardware manufacturing plants to Canon, one of its major creditors. This was the 'Steving' of the hardware I had just bought in October 1993.

NeXT now concentrated on its powerful operating system, which by now could run on other platforms, and it rebranded it as NEXTSTEP. This later became OPENSTEP, with WebObjects. Here's the graphical transition:

-->

NeXT to NeXTSTEP

To the OPENSTEP 'boxes' logo

And finally...

-->

OPENSTEP WebObjects to OSX WebObjects

Incidentally, the largest website use of WebObjects is on the iTunes store site - one of the largest traffic sites on the web.

Note a small detail on the last icon - the rivets at the corners. These are characteristic of files from what was called the 'NeXT Interface Builder' (nib) - a graphical way of designing interfaces dating from the early 90's. You simply drew what you wnated to appear, and click-and-drag connected the code to the appropriate buttons and fields. Amazing. Here is the icon from the NeXT system for .nib files:

NeXT .nib file icon

Applications in OSX are full of .nib files. NeXT lives on, albeit under cover (shades of Lt. Sullivan, a rumor monger with a column in the NeXTWorld magazine).

At some level, Jobs must relish the fact that he brought back the core NeXT OS, and replaced the Apple Classic OS - thus finally conquering Apple. It makes sense, since it really is an easier OS. But I bet he never anticipated that the item that would be the core money-maker for Apple would be the iPod.

Sigh.... Time to go into the basement. Bye-bye!

That's all she WroteNow!

Sunday, November 28, 2004

Meadows, Randers & Behrens:

The other day on NPR there was a piece about the new MOMA in New York, and how they had greatly expanded their available space to exhibit art. What struck me was the fact that they had remodeled only recently (in the '70's , I think), and that this time they had made sure they would have space for a longer time before another remodeling/expansion was needed.

It made me think that there are certain institutions or pieces of social infrastructure that are particularly vulnerable to stress from exponentially increasing populations, and others that are more adaptable, and scale well.

Examples of things that scale well fall into a class of things that are easily (and sometimes unfortunately) replicable: roads, housing tracts, and shopping malls.

Other classes of things really only can exist as single instances: subway stations, heads of state, and collections of original art. All of these are individual organizations or institutions that are supposed to serve the whole community, city, or nation. The only way yhey can continue to serve an exponentially increasing population is by increasing in size themselves (and when I refer to heads of state, I don't mean to imply that presidents become portly, but to the fact that the Executive Branch, and specifically the Executvie Office of the President has to deal with more issues with more or less the same sized staff).

I suspect that it was vastly easier to double the number of malls than to double the size of the MOMA in New York.

Tuesday, November 23, 2004

Le Blanc et le Noir:

...with apologies to Stendhal.

A sad day for me, as I let go of a big piece of my past. It is time to say good-bye to my NeXT machine. And hello to my new iMac G5.

At least the NEXTSTEP OS lives on in the Panther OS - I get to see little bits of it each day. Sometimes, the heritage is quite obvious, like alert noises, or the shake of the login screen when you get your password wrong, or the dimples for resizing. Heck, there's even NetInfo.app on both machines, for all the good it has done - I still have not been able to get the NeXT or the iMac to recognize the other machine over the Ethernet so that I can rescue files without doing the floppy disk 1.5 Mb piece by piece transfer... sigh.

What is more stunning is the difference in specifications:

	NeXT	Mac	Factor
Model	colorstation (slab)	iMac G5
CPU Type	Motorola 68040	PowerPC G5 (3.0)
CPU Speed	25 MHz	1.8 GHz	72
RAM	16 Mb	1 Gb	62.5
Bus Speed	25 MHz	600 MHz	24
Display	17 in, 1120 x 832, 12 bit colour	20 in, 1680 x 1050, 32 bit colour
Hard Disk	105 Mb	250 Gb	2380
Drives	3.5 in DD/ED FDD	CD-RW/DVD-RW
I/O	10 Mbps Ethernet, (2x)RS-432, SCSI-2, DSP port	USB, FireWire, 100Mbps Ethernet, Bluetooth, Airport, Motorola Jump

Out w/ the old, in w/ the new

Monday, November 22, 2004

San José Nights:

On my last trip, I spent a night in San José, Costa Rica in a hotel that had lots of hummingbird feeders. Except that most of the feeding occurred at night. And not by hummingbirds.

At night, a gentle flutter at the hummingbird feeder...

...led to my discovery of these fellows, about twenty of them, who emptied out the feeders in a flash. Bats. Nectar feeders, who were a little spooked by me, but quickly regained confidence as I remained frozen with outstretched camera arm on the veranda.

Bony little bugger...

Sunday, November 21, 2004

"Lime tree very pretty and the lime flower is sweet... huh?:

...apologies to P,P&M.

I know the picture below has crummy contrast, but one of the cans in it set off a cascade of memories.

Green, yellow... what's the diff?

On the left, a can of diet Coke with Lime from the USA. On the right, a can of Coca-Cola light with Lemon from Brazil.

What's the big deal, you say? "The colour of the citrus on the cans!", I yell back. They're both green! What is missing of course is the third can - diet Coke with Lemon from the USA, which would have a yellow fruit, right?

When I was growing up, we could only get small green citrus fruits, called limones in Colombia. So we called them lemons in English. There were plenty of other citrus, including limas, but they looked nothing like this. I grew up thinking a lemon was a small green citrus fruit. When I came to live in the U.S. I was greatly confused by this enormous yellow thing called, incomprehensibly, a "lemon." People thought I was an idiot. It still throws me. So there, all you unbelievers - incontrovertible proof that lemons are green.

...well, OK, in South America at least.

Oh - and the two cans taste different, too. And dagnabit, I prefer the one I can't get.

Saturday, November 20, 2004

Goscinny & Uderzo:

Well, I was supposed to post this on September 29th, but I seem to have been a bit busy lately, and have over 20 draft postings piled up...

At 13:35 GMT on that date, we had a brush with fate. You didn't feel it, did you? Don't worry, not many people did, and it's below most of our personal "worry radars" these days (since there seems to be so much clutter on our screens lately).

A 4.6 by 2.4 km hunk of rock called 4179 Toutatis whizzed by within 1.55 million kilometers of us. Here's a schematic:

Diagram by Paul Chodas (JPL/Caltech)

Now before you run outside, Chicken Little, this was Toutatis' closest approach since the 1300's, and no collision is likely for orbital projections over the next 500 years. Great, huh? "Errr.... you mean it's been this close before? ...and will be again?"

Why, yes, Chick Li, Toutatis is one of several hundred Earth Crossing Asteroids that, umm, ....well, are occasionally in the way as Earth trundles along around the Sun. Toutatis happens to be a particularly friendly fellow (in a gallic sort of way) because his orbit is almost coplanar with ours - that is good news and bad news. The good news: we get to have a good look at him every so often, and learn lots about asteroids. The bad news: one day he will be too close, and we will crush him. "We" being the Earth. If you are a mere human, well, then it might be you that is crushed. Or burned. Or drowned. Or starved.

...Chicken Little? Where are you? ...ah, there you are. That desk isn't going to give you much shelter against several hundred thousand tons of rock traveling at Mach 25, you know. There, there. Let's talk about risk. First of all, what does the picture above really mean? How far away was this bloodthirsty Toutatis thing, really?

Well, the picture shows that Toutatis' closest approach was about 4 times the Earth-Moon distance. The distance between them is about 384,000 km. Now - we can clearly see something that big. But how big are the Moon and Earth in this picture? The Earth is about 12,740 km across, while the Moon is about 3,476 km across. So the Earth would be about one thirtieth of the size of the orbit, and the Moon would be less than one-hundredth of the orbit. It's very likely that on your monitor, Chicken Little, the Moon is smaller than a single pixel. Now imagine how incredibly tiny Toutatis is - one-hundred thousandth of that orbit. About 0.1 micrometers on your screen. About the size of that bacteria's nucleus. No, the other one.

Now that you are pointing to the screen with your scraggly little toenails, Chicken Little, imagine how big your toenails are in that universe! As my mapping professors Clarence Allen and Leon Silver used to remind me mercilessly: "Always know how big your pencil is! How wide is your pencil mark in the real world?" So - your toenails are larger than Jupiter. (Here's a set of scissors, and don't leave the clippings on the rug.)

The point is that the real things we are referring to in that picture are actually very very very small compared to the distances between them. The distances between them are so large that we can't even use everyday objects to point them out properly - things like fingers, chicken toenails, and yes, even pixels, are just too huge. That means that what astronomers call a "near miss" is actually very far from being a "hit." Very far.

Let's look at some other near misses, past and future.

On March 31, 2004 FU162 missed us by between 4,000 and 11,000 km. Yikes! That's a lot closer than 1.55 million kilometers. But then FU162 is very small, only 5 to 10 meters across - small enough not to make it to the surface, but it would have been LOUD. We actually had a closer encounter than that - on August 10, 1972, something about that size or bigger missed us by only 53 km. more

In the "big hitter" category we have 2004 VW14, which on December 24, 2004 will pass within 1.8 million km (about 5 orbit radii away). Current interest centers around 1997 XR2 (0.23 km diam) and its approach on June 1, 2101 about a 1 in 10,000 chance of collision, or about 99.9903% likelihood of a miss. A little further on, on August 24, 2130, body 2004 QA22 will pass by between zero and 653,000 km away. Zero, of course, is a collision. This by no means should be entered in your calendar as the apocalypse - the orbital parameters for these bodies are updated frequently, changing these estimates by large amounts.

The only body for which a good degree of confidence exists for collision is 1950 DA, with a date of March 16, 2880. However the collision probablility is still only between 0 and 0.33%. (a 99.68% chance it will miss) more

Here is a decent animation of how we reduce uncertainties for asteroid postions with repeated observations.

JPL page w/ lots of Toutatis data: http://echo.jpl.nasa.gov/asteroids/4179_Toutatis/toutatis.html

A web page by one of Toutatis' co-discoverers, A. Maury.

JPL Close Approach page

So, my little C.L., I hope this calmed you down - you can stop shedding and sh###ing all over the carpet now. Just watch out for the cats.

Friday, October 29, 2004

Jenner, Pasteur:

Hearing so much influenza vaccine hype surrounding the election just makes me angry. Where are the facts?

Yes, we are short of our usual supply. But what does the 'usual supply' tell us about participation rates in the vaccination program? The usual supply is about 100 million doses, which means that at best the usual maximum expected inoculation rate for the U.S. population is about 33 percent. That's not a very good rate.

Even if we looked at the most vulnerable group, those over age 65, the historical inoculation rates are poor. The average is about a 66% inoculation rate for the nation's 36 million seniors. If we turn to the other risk groups, the rates fall below 50% - those with chronic conditions, pregnant women, and infants under 6 months. Historically we could expect that about 40% or 40 million of the doses would go to these high-risk groups. The sad fact is that we have never yet met targets for inoculating this group.

We will have 100 million doses by mid-January, enough to cover the historical rate of inoculation. However, the panic caused by the incessant media coverage of the shortage and its political fallout has caused an increased demand. So, once the newly arrived doses arrive and are administered, we will actually have inoculated more people than usual, especially in the high risk groups. So that's actually good, right?

Well, of course it's more complicated than that. It depends on when the flu season peaks, and that depends on what strain dominates. Since the flu virus is so adaptable, the vaccine has to be reformulated each year. And since it takes time to gear up the production, the decision on what strain to put in the vaccine has to be taken in February of the previous flu season. If the guess is wrong on what strain will dominate, as it was in 1999-2000 and in 2003-2004, the flu outbreak will be more severe than usual.

Here's a nice graph of flu case numbers in a Baltimore hospital over successive seasons from Johns Hopkins:

In 2000-2001 and 2001-2002, flu cases peaked during the third week of January, but in 2003-2004 they peaked much earlier, in the second week of December. If that happens this season, you can see that there will be a lot of fallout, because the additional doses of vaccine will only be available after the peak.

A note of caution here though - there is no clear statistical correlation between rates of vaccination and hospitalization rates for the flu. In other words, historically the vaccinations have made no difference in how many people are hospitalized. The other point is that when a new strain appears, it doesn't matter whether you were inoculated or not, since the vaccine will not have that new strain as part of its make-up, and you will be vulnerable. Against a new strain, there is no defence. The CDC estimates that between 15 and 60 million U.S. residents will get the flu in an average year - this is an estimate since most people don't report flu to their doctors, and even fewer report flu to hospitals, the source of the firmest statistics. Of these, between 40,000 and 55,000 people will die from the flu - but this is a very fuzzy figure, since it is derived from death certificates, which are notoriously inaccurate.

My conclusion - if a new strain breaks out and this is a bad flu season, there will be a political firestorm. And no-one will pay attention to the fact that this would have occurred even if we had had the regular supply, and even if we had the unprecedented situation of 100% inoculation.

Who cares about the facts when there is political hay to be made?

Thursday, October 28, 2004

Hummer H2? Ho-hum!

I read yesterday that Navistar International has released an SUV larger than the Hummer H2. Much larger. The International 7300 CXT is for you if your weekend "to do" list includes hauling up to six tons with a 220 HP diesel engine that gets about 8 mpg. Lucky that comes with a 70 gallon fuel tank. You might have to get a special permit from your DMV for the airbrakes in order to drive it. You'll also have to pay over $250 per tire should you damage the giant 11R22.5 tires while off roading with your 8-foot pickup bed full of whatever it is your spouse wanted you to haul. I have no idea what the rims cost on these beasts.

But I have to admit, it is pretty. If you like trucks. Big trucks. And this is a BIG TRUCK:

Photo (c) Navistar International Corp., 2004

The release tells me that the industry is confident that those American consumers who can afford the $95,000 base price are not deterred by $2/gallon prices. There's even one on eBay today with a starting bid of $99,900 - with 7 hours to go on the auction, it still has no bids, but being eBay, I'm confident that if it moves, it will all happen in the last few minutes...

In other news (and perhaps compensatory, in emissions) is an H2 Hummer converted to run on hydrogen, bio-diesel, vegetable oil, or compressed natural gas. Amazing - that way you don't have to worry about where the next hydrogen filling station is, as the straight, boring H2H hydrogen-only powered Hummer would have to do...

Of course the key here is that Navistar is cranking out the 7300 CXT on a production line, while these modified Hummers are both one-off jobs. There's little compensation for the emissions these mammoth vehicles will be putting out. And that's not even considering that to produce hydrogen fuel, you still have to consume energy that comes from... you guessed it, more conventional fuel than the hydrogen saves. As of right now, hydrogen is still actually a dirty technology.

Friday, September 17, 2004

Kansas, I don't think we are in toto any more...:

An old MIT friend now directs a program that takes Doppler radars into as direct a line with trouble as possible. Josh Wurman and his team recorded the highest near-surface windspeed ever, 301 mph in a tornado with equipment like this.

This is a MODIS shot of Frances, just off the northern shore of the Caribbean island of Hispaniola. Frances is in the middle of destroying the Haitian city of Gonaives - relentless denudation of the hillslopes for firewood above the city left no barriers to stop the topsoil from being washed away into torrents of mud.