Cats and Supercomputers

There is probably a reason most facilities that host Beowulf class cluster computers or supercomputers don’t also host cats.  One week of cat hair, especially during the Spring Shed, can be a real problem … also, vacuuming out the fan intakes can cause fan speed changes that sets off all kinds of interesting and entertaining alarms.

While keeping an eye on Kilauea and other geophysical hazards, tropics are quiet at the moment although some potential areas in the West Pacific …

Alberto remnants from GOES-East

Very pretty swirl of clouds – not so pretty flooding in Central Alabama (parts of which are under a “Flash Flood Emergency” warning), North Carolina, and there is still a lot tropical moisture being pulled up into the rest of the South East … click to embiggen.

Admin note: also testing to see what it takes to get FB and Twitter to embed images … yes, I can write climate models from scratch and solve diffy-q’s in my head, but social media is still a pain sometimes!

#Alberto related flooding

Although Alberto isn’t being formally tracked as a tropical system any more by the hurricane center, it actually has had a better radar presentation as a swirl of rain than it ever did as a storm!  Unfortunately the “remnants” are still causing problems across the southeast and now central US.  Here is the radar and warning areas from just before 8am Wednesday morning, showing flood watches and flash flood warnings across Alabama, North Carolina, and Kentucky … click to pop open a larger version.  Given how saturated the ground is, and all the tropical moisture streaming into the region, folks across the Southeast and central US who live in flood prone areas should pay attention to local warnings and be prepared.

What do storms “cost”? A look at #Alberto

The National Hurricane Center has issued its last forecast for Subtropical thingee Alberto.  Here is the Svarog/Perun model impact estimate based on the 5am Tuesday 29 May track.  Also noted is the forecast track from the 26th.  Despite the fact this was a disorganized blob that wasn’t very tropical, and “reformed” several times, NHC did a good job on the track.  Their intensity estimate was too high, the original thinking is Alberto would be just below hurricane strength at landfall whereas it was barely tropical storm force, but given the holiday weekend, and the real potential for strengthening if the storm transitioned to a more tropical structure, the NHC forecast from Friday was not unrealistic.

So how much does a storm like Alberto “cost?”  That’s a tough question, and to some extent a matter of perspective, so this is a long post but hopefully interesting.  Unfortunately insurance industry terminology drives much of the discussion around natural disasters.  So the term “loss” is commonly used.  But that is misleading – certainly it’s a loss to the insurance company bottom line (although paying claims is sort of why they exist!), but it’s a “gain” for the contractor who gets the work to do the repair.  It’s a partial reimbursement for the homeowner or business, because of course there is the deductible and expenses that are not covered, much less costs to governments, the uninsured, and so forth. This is further complicated by the fact that flood “losses” are not covered by private insurance at all in the US, rather, it is covered by the Federal Government’s National Flood Insurance Program (NFIP).  Specialized coverage like business interruption insurance is complicated, and not often carried by small businesses.  So more often than not, insurances “losses” are actually not reflective of the total impact of a storm.  In fact, since the mid 1990’s (after the shock to the industry of hurricanes Hugo and Andrew, as well as earthquakes in California), deductibles and exclusions have risen dramatically, and the total share of the impact of a storm borne by the insurance industry has dropped.  The industry covered about 60% of the costs of storms like Andrew, Hugo, and if you include NFIP, even Katrina was over 50%.  But the coverage of recent major storms has been between 20 and 40 percent, and for weaker storms 10% or less.  We think that the amount covered by all forms of insurance of last year’s Harvey was less than 30%, and for Irma not much better.

What about just using repair and replacement costs, irrespective of insurance?  Well, even that is complicated.  Take the case of a small beach home build in the 1960’s.  If it suffers more than 50% damage, it cannot be rebuilt as it was, rather, it must be rebuilt to modern regulatory standards – that means it must be elevated, higher wind standards, and so forth.  So that $150,000 cottage is replaced by a $400,000 structure, or worse it and its neighbors are replaced by a multi-million dollar condominium. So what did the storm “cost,” $150k, or $400k, or more?  Depends on your perspective.  And of course that doesn’t include lost wages, revenues, and other economic impacts.

As a side note, we have seen this happen across the coastline in the southeast, that after a storm inexpensive beach cottage communities are replaced by condos and McMansions.  This is an unintended consequence of forcing higher construction standards and mandatory flood elevations on the immediate coast: while it might make structures more resilient, it also drastically increases the value of property at risk. The net effect is to actually increase storm costs!

As noted above, one persons “loss” is another persons “gain.”  Strictly speaking, unless your bars of gold are buried and you lost your map, most storm impacts are not “losses” in a macroeconomic sense, but redirected spending.  Instead of taking that vacation in the south of France, your insurance executives are having to slum it on Waikiki 😛  and your local roofing contractor gets to hire some part time help and buy a new boat 🙂 OK, that’s a little jaded, but you get the point.  What natural disasters do is change where the money flows and if/how/where it is invested, not the total amount of money in the system except in rare cases when government print money to increase spending (which has other negative impacts).  This leads some economists to talk about a “disaster stimulus”.  That is wrong in my view for a variety of technical reasons that only an economist would stay awake to hear, but in simple terms, storms take the intelligence out of the system and forces spending in ways that would not occur otherwise.  For example, instead of expanding my business, which would improve my community, I may be spending the same amount of money (or likely more) to just get back where I was.

So what exactly are we at Enki estimating, and how do we come up with an estimate?  Our goal is to estimate the economic impact of the storm.  By “economic impact,” we mean the value of redirected spending that is not recovered by short term displacement.  For example, grocery stores often see a ‘bump’ just before as storm as we stock up on Cheetos and Mountain Dew (hey, we’re scientists here; good nutrition is for sissies).  During the storm they see a “loss”, and see a bump afterwards as people buy stuff they delayed getting due to bad weather.  For the oil and gas industry, we don’t count the cost of “shut in” production unless it is not compensated for via increased production after the storm passes or from other sources.  So you can see it’s already not simple.  We start with physical damage and replacement costs.  That is somewhat straightforward – the computer models that estimate such things are actually pretty good.  We add in physical costs of infrastructure – stuff that tends to be overlooked, such as replacing signs, street lights, and so forth that local governments are often stuck with.  Cleanup costs are included, and can be a significant expense to both government and the private sector.  We then delve in to things like permanently lost wages (which hits hourly workers especially hard), and service industries.  Extra transport costs, and so forth come in to play.  The models are thousands of lines long and even take in to account the time of year and days of week of the impacts – very important for a storm like Alberto.

So given all that, our models are currently estimating the economic impact of Alberto to be somewhere around $700 to $800 Million.  Insurance based estimates of the “cost” of Alberto will undoubtably be small.  With high deductibles, most of the impact due to lost revenues or flood damage, the insurance industry share will certainly be less than 10%.  NFIP will get a bit more, but the vast majority of the impacts will be to small businesses that took a hit in the form of decreased income over the holiday weekend.  For comparison, we ran the same scenario two weeks earlier, before school is out, and the impacts were under $150 Million.  Two weeks later (which is in to the summer Travel season in Florida) and it is nearly $400 Million.  So for a tourist laden area like Florida, timing is everything …

Swirl-o-clouds Alberto makes landfall

It looks like Subtropical Storm Alberto has made “landfall” between Panama City and Pensacola after skimming the panhandle of Florida. Here’s the radar and 4:40pm Monday, note the flash flood warnings (pink) as well as some severe thunderstorm warnings up in Georgia near and north of Waycross:

Not much change to report other than the collapse of the rain over the rest of Florida.  Still some risk of brief tornadoes in stronger lines of thunderstorms (like up in Georgia), and flash flooding is the big potential risk.  As noted this morning, the likely physical damage from this storm is in the low hundreds of millions at the very worst – but the disruption to the holiday weekend will be felt by small businesses across Florida, Georgia, and Alabama.  Winds are “only” around 45mph in the storm cells, so that means branches down, maybe a stray tree or two, that sort of thing.  Another mass of wet air is lurking across Cuba and is likely to move north, and the entire southeast is covered by a very wet air mass, so more rain is likely across the region.  So if you are in a place where a couple inches of rain can cause flooding, beware!


Alberto slows down, further west, weaker; impact forecast now $800 Million

Alberto pulled some dry air into its circulation last night, and slowed down, so it is not making landfall early this morning.  It is also weaker than forecast.  Here’s a 6am satellite image with the RAP13 model pressure field.  This is using a wavelength of light known as the infrared water vapor band.  It lets us look at the amount of water in the air.  You can  see a band of blue and white wrapping around the storm, the moist air that is the “fuel” for a tropical system.  But you can also clearly see the orange/yellow dry air to the west, and a band of dry air that is “entrained” and wrapping around the east side …

The radar shows the same thing: although there were heavy rains overnight, it has stopped raining here in coastal Georgia …

The 5am National Hurricane Center forecast reflects these changes and now has landfall at late tonight, further west than yesterday evening’s forecast, and much weaker. So our damage forecast has dropped some, but not as much as you might think.  Most of the economic damage from Alberto has already been done, in the form of a disrupted Memorial Day weekend.  So the estimate of $1.1 Billion only dropped to $800 to $900 Billion – at least $600 Million of economic impact is already “locked in” even though the storm hasn’t even hit land yet!

Here’s the current (5am Monday morning) forecast and impact map …

Alberto Sunday Evening Update

Not expecting any major changes in the 5pm NHC package.  Rain associated with Alberto had now made its way well in to South Carolina as well as covering southern Georgia and all of Florida:

“Landfall” should be in the morning, but given the wide, diffuse nature of the system, other than for academic purposes it might be hard to tell exactly when and where unless it gets a sudden round intensification or organization, so the impact estimates haven’t changed since this morning.  Flooding continues to be the main worried, some spotty wind damage, and there is some potential for an isolated tornado.

This is neat – it’s an animation of 30 minutes of the “mesoscale” mode scans from GOES East … it’s big so it may take a bit to load before it starts to loop.

Alberto Mid-day Sunday, 27 May

Not that much has changed with the 11am NHC forecast update, with landfall still showing over the Florida Panhandle on Monday Morning, with peak winds (offshore!) of 60mph; over land will be less.  Lots of rain over Florida, now moving in to far southern Georgia with more offshore.  An Air Force Hurricane Hunter found the interior structure is improving, with a shallow warm core (a key element defining a tropical storm) now evident.  So the story is that flooding is the main risk from Alberto, some on the immediate coast, but mostly inland in flood prone areas that are already saturated.

With a broad wind field, hazardous storm surge is always a concern even far from the storm center.  Here is the animation from NCEP’s Extratropical Storm Surge model.  While it’s not really a “Tropical Cyclone Storm Surge Model”, it will probably do a better job on this storm because Alberto will still more resemble an extratropical storm at landfall than a traditional tropical storm. Peak water levels in the Big Bend area of Florida could run over a meter (3.5ft) above normal.  Click to load, it’s sort of big so be patient.

Finally, here is an animation from GOES-East of sunrise over the storm.  Note the improved structure as clouds wrap around the center … click to embiggen/animate.

Economic loss estimate didn’t change much, still around $1.1 to $1.2 Billion.  And as noted, much of that will be “secondary” impacts rather than direct damage.  No substantive risk to the offshore Oil and Gas infrastructure, some temporary disruptions due to shut in production and tanker traffic being delayed, but nothing serious.

Alberto Sunday Morning Update; Economic impact forecast $1.1 Billion

Alberto is starting to organize a bit, both on satellite and now radar.  Here’s some views from early Sunday morning (5-6am).  First, the infrared satellite, with the estimated surface pressure isobars based on the NCEP “RAP” model (which integrates data and does a short term forecast every hour) – you can click any image to see at full size …

Here is the radar composite view at 6:45am …

And finally here is the forecasted impacts from my Svarog/Perun model, using the 5am Official Forecast Track from the National Hurricane Center.  Also shown is the previous (11am) track – if you’ve been following this, notice they have shifted the forecast eastward again.  This isn’t so much because the “storm” is deliberately moving that way, but that the center has up till now been somewhat ill defined and keeps reforming.  As you can see on the radar and satellite image, the system is starting to organize a bit better and develop a more traditional tropical structure.  I expect it will lose the “subtropical” and moniker and become “tropical” late today.

If this track and forecast holds the main impacts from the storm will be rain over a very wide area including all of Florida, Georgia, and Alabama, as well as parts of adjacent states , but gusty winds and some shallow coastal flooding are likely along the immediate path was well. If you are in the path of the storm, take reasonable precautions, especially if in a low lying area that can flood, or a vulnerable structure like a mobile home.  If you include the costs of disrupted business, government preparations, and so forth, Alberto is shaping up to be a $1 Billion dollar storm.  That’s not a lot these days, and in perspective of the $16 Trillion overall US Economy, no big deal.  Not much consolation if it is your house that is flooded, or a tree falls on your car, but … to put it another way, Alberto likely to cost the equivalent of 48 days of potato chip sales 🙂

Economics Addendum:  It is highly unlikely this storm will have any substantive impact on oil and gas production in the Gulf.  It’s not that strong, it’s not traversing the main fields, so aside from some precautionary shut ins and tanker traffic delays, unless something breaks that shouldn’t Alberto won’t have any impact on energy prices other than giving traders a chance to play with the numbers a little.

Alberto Saturday Evening 26 May Update

As the sun sets over the western Caribbean, Alberto is starting to develop a swirl of clouds, but still isn’t much more than a big “comma” with most of the clouds and rain many hundreds of miles from the center, which is located just off the tip of Cuba … this is (hopefully!) an animation of the storm from GOES-16 this afternoon that you can click to get high res:

As expected NHC continues to shift the forecast track eastward as the storm formation location has moved and the driving forces seem to be pushing it more that way.  Here’s the 5pm EST track and impact forecast from my Svarog/Perun model, based on the official NHC forecast.  While the landfall location hasn’t shifted much, they are showing the storm center shifting much closer to the west Florida coast.  Since the worst impacts are normally on the east side of the storm, that means worse conditions over the state.

The main risk from Alberto no matter if it technically becomes a tropical storm or not, is rain and flooding.  The Southeastern US is saturated from over a week of wet weather, and another 4-8 inches doesn’t really have anywhere to go.  That said, saturated soils make it easier for trees to blow over, so that’s not so great either even if the winds are not that strong. Here’s the radar composite from just before 5pm …

I hope this new format works better for everyone  It’s certainly much easier for me, and I think lets me put more stuff in (like the higher res animations) that FB just doesn’t seem to permit.