Tropics June 3rd, 2020 (TS Cristobal and Cyclone Nisarga)

Two storms making landfall on opposite sides of the world this morning.  Cyclone Nisarga has hit the west coast of India, south of Mumbai with hurricane force winds.  It rapidly intensified from a weak depression to full blown hurricane in less than a day.  The impacts both financial and virological are likely to be significant but hopefully not catastrophic. Here’s the damage swath …

In the southern Gulf of Mexico, Tropical Storm Cristobal is hovering near shore, dumping a lot of rain on the Yucatan peninsula, causing mudslides and flash flooding.  It will likely drift inland and start to break up, however, in two days whatever survives is likely to be dragged back offshore and northward towards the Louisiana coast.  If that scenario holds up, as the NHC is forecasting, the impact swath will look something like this:

The main uncertainty is how organized the storm will be as it begins its northward trek.  The more organized, the stronger it will be when it hits the US.  Likewise, it may be nothing more than wind and rain.  We will know more by Friday morning …

 

Pointing the finger

On the doom front, a few earthquakes, no tropical systems at the moment. Nothing different on the pandemic data front: it’s still so screwed up it is hard to draw any conclusions.  Maybe things are getting better.  Maybe we are in the eye of the storm, and a “second wave” is building.  We can’t know that yet here in the US because the data collection and testing is so screwed up. But that isn’t stopping the finger pointing. Bloomberg ran an editorial yesterday entitled “The Pandemic Is Exposing The Limits Of Science.”  The author, Ferdinando Giugliano, tries to draw parallels between the 2008 financial crisis and the SARS-COV-2 pandemic by asserting that the 2008 crisis showed the limits of economics in understanding the economy, and now he asserts the same regarding science and COVID-19.  The editorial entirely misses the point in both cases.

For years, economists had been warning politicians (and business journalists) that there was a coming storm.  Assets were over-priced, there was a well known real estate bubble, and leveraging had created an environment where several major financial firms were exposed to collapse, putting the banking system at risk. Business journalists pooh-poohed the “doomsayers,” and politicians cashed the checks of the financial services lobbyists and smiled.   While it is true nobody knew exactly when the crisis would come, saying “nobody warned us that in Summer of 2008 it would hit” is like saying “well, you may have warned me driving 200 mph was dangerous but you didn’t warn me about the oil patch that caused me to skid off the road and hit a tree at 4:35 Saturday Afternoon.”  That is absurd.

Giugliano writes “But on a range of issues — from containing the virus to prescribing effective treatments — we have seen some scientists and doctors jump to conclusions, only for others to give immediate rebuttals. (The contention over the efficacy of hydroxychloroquine is one example.) This seesawing has added to the sense of panic and confusion among ordinary citizens.”  That is infuriating.  The main reasons these became issues is that journalists failed to do proper reporting but in their constant need for “breaking” news it is “reporters” who jumped to conclusions, publicizing preliminary results that had not been peer reviewed.  Scientists are in a catch-22: if we try to keep preliminary results confidential we are accused of a lack of transparency and hiding things from the public.  If we are transparent, reporters grab and publicize raw results and misinform both the public and political readers who don’t read primary publications.  The Hydroxychloroquine issue wasn’t pushed by scientists or doctors.  It was pushed by politicians who were repeating irresponsible “news” articles based on raw, preliminary research.

So I would like to point the finger too.  The middle one.  At journalists and politicians of both parties who can’t be bothered to learn how science actually works, won’t listen to neutral subject area experts, then cherry pick and dramatize raw data and exaggerate uncertainty for their own agendas, placing the blame for a crisis elsewhere instead of where it belongs: on them.

Why Data Matters: Hurricane Season Forecast and COVID19 Numbers (again!)

Where to start, the screwed up COVID19 testing/hospitalization data, or the much more useful hurricane season outlook (it’s an outlook, not a forecast, for technical reasons)?  The hurricane outlook is far more useful and far more scientifically sound, but it does suffers from a similar problem to the COVID19 data: inconsistent standards over time.  (Of course, the COVID19 data suffers from many more aliments).

Here’s a summary of  the 2020 Atlantic Hurricane Season Outlook from NOAA:

The full release is here.  My quibble with it is that the seasonal statistics it is based on were not compiled in the same way, with the same criteria, that the outlook is.  Saying a season is “above normal” in comparison to seasons in the past, especially before satellites (the release also talks about some of the exciting new data and models coming online this year) is questionable.  As we have better data, the standards and thresholds for what is a tropical storm, hurricane, and especially intensity estimates have gotten a lot more refined.  I never liked storm count scoreboards for that reason.  In any event, what matters is that the ocean and atmosphere are primed for a busy season.  Of course, the total number of storms is irrelevant if it hits you.   So if you live on the coast, prepare.

What can be said about the COVID19 data that hasn’t been ranted before?  Are things getting better or worse in the states that are reopening?  We just don’t know.  Obviously things aren’t turning into a New York style dystopia, but the testing and reporting are so bad and so inconsistent that as noted before we have no idea what the hell is going on (sorry, but it’s time to roll out the profanity).  Lumping in antibody and live virus tests as apparently several states are doing, taking weeks to report hospitalization data, it’s all so screwed up you can make any argument you want based on the numbers floating around.  Or just roll dice and throw darts.  Here’s Georgia’s aggregate hospitalization data, and reported persons in the hospital, for the last five days:

If we had real time reporting, these would match.  If there were a one or two day delay, you would see a time lag but the numbers would match with the number of days delay (in other words, the Thursday aggregate change would be seen in the Tuesday hospitalized change, if there were a two day delay).  These are just different data sets, and I don’t believe either one.

So what does this mean for you?  Try to stay safe.  Hygiene, distancing where possible, masks where not.  Take special precautions for those over 65, and anyone with health problems.  Feel bad stay home.  Unfortunately the CDC main COVID19 site hasn’t been updated since April 19th.  Some reopening guidelines are here.  We’ll see what happens … maybe it will be ok, maybe it won’t.  Without good data, there’s no way to know if it isn’t until it’s too late.  Likewise, people have no confidence to go forward if it is in fact ok, which makes the already catastrophic economic damage worse.  So the confused data and monitoring are hurting either way, and it shouldn’t be a political issue.

Amphan and COVID19 Updates (Georgia Data again)

Tropics: Amphan is approaching landfall on the Indian/Bangladeshi border, unfortunately a bit stronger than forecast yesterday.  It is now on track to cause over $1 Billion in impacts in this impoverished region.  16 Million people are in the hurricane force wind swath, and 1.6 Million are at risk of severe storm surge and river flooding.  Arthur is no more, a cold front from Canada did him in.  Mynd you, møøse bites Kan be pretti nasti.

Pandemic: The Georgia numbers are contradictory.  If you want to be optimistic about the re-opening, you can look at the report that hospitalized COVID patients decreased from 1025 to 986, a decrease of 39.  Or, if you want to be pessimistic, you can look at the total reported hospitalizations, which increased from 6916 to 7027, or an increase of 111. Of course, these two metrics are measuring different things, but in a  system with reasonable reporting and time lags they would agree more closely.  How anybody can look at this data and say we know WTF is going on in this state is delusional.  Mortality rate, fraction tested, hospitalized ratios, etc. all are moving slowly, oscillating within the noise levels with no clear signal.  The jury is still out on the reopening.  Some have asked about county level data.  In my not so humble opinion, it’s just not worth bothering with.  It’s too noisy, and I don’t trust it.

Elsewhere, most states other than New Jersey (which is still in a growth curve) are in a slow growth curve.  Here’s the latest for several US States and the composite US (with and without NY/NJ):

In the rest of the world, Spain, Italy, and while lagging in time, the UK, are all coverging towards mortality rates of 5.5 to 6.5 deaths per 10,000 population.  France, Netherlands, and Sweden are headed to the 3.5 to 4.5 range.  Russia now seems to be following a track more like Sweden than Canada, and will likely end up in that same range.  Canada will like be around 2/10k.  The bottom tier of South Korea, Norway, Germany, and Denmark are in the 1.5/10k range or less.  For context, the H3N2 influenza outbreak of 2016 had a rate of just under 3 per 10,000.  As a whole, the US is on track for under 1.0 – but as seen in the above graph, it depends a lot on who you are (old/infirm vs. younger/fit) and where you live (NE vs rest of the country).  Here’s the world plot …

So what does this mean to you?  Same as it ever was.  Hygiene.  Keep your distance from people you don’t live with.  Masks if you can’t. Sgt. Apone rules: (nobody touch nothin’, don’t bunch up).  Hopefully things are on a good track.  But given the mess that is testing and data reporting, we just don’t know.  And that is an outrage.

Reopening Georgia: What the data (doesn’t) show

This post should be short.  Instead of trying to figure out what the crappy data indicates, it should be “the data says X so we should do Y.”  While there are a couple of brighter spots, the overall COVID-19 testing, data collection, and reporting in this country continue to be nothing less than a MAJOR EMBARRASSMENT.  Red states, Blues states, forget the tribal poo-flinging.  Perhaps one party is worse than the other, but it’s the difference in degree, not in outcome: both parties have failed. So instead of having good information to make decisions with, we spend lots of time arguing over the “data.”  And if everyone wants to be intellectually honest, based on the “data” we have, you could in fact argue for a range of options from maintaining strict mitigation measures all the way to “why bother, back to normal.”  Which suits the political leadership just fine: if the data were clear, which policies would work best might be clear and they might have to agree on something, rather presenting stark contrasts which keep you angry and divided.

Let’s look at Georgia, since it was the first to “reopen”, and selfishly because I live here.  So instead of having good information on which to base the reopening decision, and thence if the reopening decision was sound, we have to read entrails and cast runes to try to figure things out.  But what it shows isn’t very comforting.  Right now I’m focusing on the reported number of hospitalizations.  Georgia’s data utterly inconsistent.  They are reporting 15 new COVIDI19 hospitalizations yesterday.  But the *total* number of hospitalizations went up by 129.  Factor of 10 discrepancy.  Which is right?  Probably neither one.  So let’s get another thing clear: reporting delays on top of the way the virus progresses through the population, means that trying to figure out the consequences of decisions made two weeks ago is frustrating at best, and potentially a less a waste of time.  But … let’s give it a shot anyway.

Here is a plot of the changes in reported hospitalizations sorted by day of week.  The “X” axis is weeks since March 19th (which were all zeros). The “Y” axis is the change in the total hospitalizations.  Why by day of week?  Because for a variety of reasons of both behavior and reporting, there is a very clear bias in reporting between the days of the week.  So you can’t compare Tuesday to Monday, much less a weekday like Monday to Sunday.  So here I’m just comparing Sundays to previous Sundays, Mondays to previous Mondays, etc.  In theory, if the reopening is causing a spike in dangerous infections (those requiring significant medical care resulting in hospitalization), it should show up in this data.

What we are seeing isn’t comforting.  We have three straight days of relative increases.  And 4 of the last 5 days saw above normal increases (Friday was a draw).  Conclusions?  The trends don’t look good, but again this is very noisy data, so we will have to continue to watch carefully.

What does it mean to you?  If you are re-engaging in everyday life, please be careful.  Maintain your distance where possible, wear a mask when you cant (for others more than yourself).  If you feel any symptoms stay home.  Good hand hygiene is absolutely essential. Minimize contacts with surfaces.   If you go out shopping, please please please DON’T PICK UP EVERY ITEM IN THE BIN!  I’m looking at you, idiot who picked up and examined every bag of mixed veggies at Publix last week!

Hurricane Planning during a Pandemic: A Dangerous Combination

There are currently three tropical systems stalking the earth: Typhoon Vongfong has just swept through the Philippines, causing significant but not catastrophic damage.  There are two “invest” (potential storms), one off Florida, and one in the Bay of Bengal (as of noon, a tropical depression being tracked as IO012020).  Here is the “big picture” for this morning … of these, the one to watch is the one in the Bay of Bengal; it has the potential to be deadly if it hits the highly populated coasts of West Bengal and Bangladesh.

Hurricane planning presents difficult issues in the best of times.  While Emergency Managers and politicians don’t like to think of it this way, and often simplistically (and wrongly in my view) present it to the public as a “err on the side of caution with few consequences” kind of thing, in fact when you evacuate an area you are disrupting society, and that carries with it both economic and human life consequences, especially for the elderly, infirm, and at the lower end of the economic spectrum for whom there is little safety net.  If you dig into the mortality statistics, every evacuation has killed people.  The gamble is that you are saving more people than you are killing.

In recent years, with the move away from a “civil defense” approach to emergency management and towards a more “law enforcement” based approach, larger and larger populations are included in evacuations.  I’ve discussed this before: the current philosophy in the US is that you want to get all “non essential” people out of disaster areas until basic services are restored (electricity, streets clear, internet and cable TV back on 😛 ).  In other words, to be blunt, the risk threshold has been changed from “life and limb” to “irritable and inconvenienced.” So emergency managers have fallen in to the habit of pretty widespread evacuations.  I think this is a bad thing in general, but that’s a longer discussion.  In the world with the SARS-COV-2 virus running around, whether that is a good approach or not is irrelevant: it has become potentially deadly.

In a pandemic, you want individuals to severely limit contact with people outside their immediate circle, to avoid spreading the disease.  This is the basic concept behind “social distancing” and limiting travel:  every additional person you come in to close contact with, be it from breathing the same air or touching the same surface, you increase your chances of getting – and spreading – the virus.  This is especially true for travel.  You really don’t want people traveling outside their immediate communities.  Even one person can cause an explosive outbreak, as was seen in the Albany Georgia area.  So you can see how an evacuation is absolutely incompatible with trying to keep a pandemic under control.  Scattering people across the region (and in fact country), traveling by car (which means rest room visits, stopping to eat, and so forth, in multiple locations) staying in crowded shelters or hotels, then bringing them back together a few days or week later, is probably the worst possible thing you could do.

So, what should emergency managers do, and what should the general public do?  My suggestion is to go back to basics.  What is your risk from physical harm?  And for that, we need to go back to the basic rule of thumb with respect to hurricanes: the majority of deaths are from storm surge and inland riverine flooding.  Especially for weaker storms (Cat 1 or 2), wind is not such a direct threat to life if you live in a reasonably well buillt home (although having a tree fall on your house is terrifying, and potentially deadly, we’re talking about overall statistics here).  So the cardinal rule is “evacuate from water, shelter from wind” with the caveat that for mobile homes, almost any winds above tropical storm strength are potentially deadly, so they need to seek shelter.

And this is where additional planning needs to take place.  In recent years there has been a trend in coastal counties to not open local shelters.  This should change.  Now.  There is absolutely no reason why, in Chatham County GA for example, there should not be shelters in-county for category 1 and tropical storm purposes. Yes, absolutely, we need to get people off the islands and out of mobile homes.  But sending them far inland is a problem.  In a COVID19 world this becomes more critical, as we really don’t want people having to travel great distances to crowded shelters and exposing themselves to people outside the community – both for their safety and ours.

For Category 2 it becomes a bit dicey but evacuating the county makes sense, and for a direct Category 3 landfall potential clearly the entire county should go, even in a potential pandemic.  Bypassing storms such as we have had in recent years are always tricky.  In all cases, however, the focus needs to be on the potential conditions in the county rather than the conditions in the storm.  Just because it is a Category 2 hurricane offshore doesn’t mean it represents a Category 2 threat to the county.

Special needs populations like the elderly, those with immune problems or other physical issues like needing oxygen, etc., have special considerations.   But, again, evacuate from water, shelter from wind is the cardinal rule, but with the twist that you need to be aware of the potential for longer term power outages as being a threat to life.  If you are in a vulnerable location, by all means get out. In that case, if you have to evacuate, again try to stay in a known environment close to home.  But the normal, usually minimized risks of stress and travel for the elderly are now compounded by the risk from COVID-19, moving the needle a bit more towards staying (if in a secure structure with supplies, considering power needs).

So the basic rules are: if in a flood zone, get out; the risk from the storm is greater than the risk from the virus.  If in a mobile home, same advice.  If in a reasonably well constructed house outside the flood zone, shelter in place up to Category two winds.  By the way, now is the time to do some limb trimming, and if there is a dead/diseased tree nearby, think about that.  Also pay particular attention to cleaning up potential sources of wind blown debris – lawn furniture, that junk in your backyard you’ve been meaning to get rid of, etc.  Don’t wait to the last minute.

In most cases following the advice of your local emergency managers is the best thing to do. Some (hopefully all) are rethinking their plans, such as those in Florida, are publicly and proactively modifying their plans.   Hurricane planning is unpleasant to think about in most years, but the COVID-19 outbreak makes it even more critical that you assess your individual situation and risks, and have a plan as to where you will go and what you will do.   And figure out how much of your carefully hoarded toilet paper to take with you 😛 !

What the Data Shows, viral and cyclonic (13 May 2020)

Lots of COVID charts today, but sadly no math 🙁 .  First the tropics: The data shows that Typhoon Vongfong will sweep through the Northern Philippines over the next few days.  The bad news is that it is strengthening faster than anticipated and will be much stronger than forecast yesterday, but the good news is the track has shifted so that the brunt of the storm will stay offshore.  Here’s the map:

On the pandemic front, nothing much has changed.  While testing has increased, it isn’t really changing the picture much, which seems to indicate the virus is spreading about as fast as the testing – fortunately (or not depending on your perspective), neither very fast.  Yes, the fraction of people tested is increasing …

BUT, the fraction of those tested showing up positive, and the fraction of people tested who are in the hospital, remains fairly constant, especially the ratio of hospitalized to tested, which is an amazingly flat rate for the areas we have data: Continue reading

Potential typhoon headed to the Philippines; impact on pandemic

For those who are tired of the pandemic posts, it’s that time again … we have a storm in the West Pacific.  But even here the shadow of the SARS-COV-2 virus can be seen. The Joint Typhoon Warning Center (JTWC, who are responsible for tracking for the US in that region) has started formal advisories on WP012020. WP01 is being called Ambo already by the Philippines weather service, who have their own naming convention; if it strengthens as forecast it will be named  Vongfong by other agencies like JTWC).  The storm is forecast to become a weak category one hurricane (typhoon) just before landfall, and sweep across the central and northern Philippines over the next five days.  While not particularly strong, the winds and rain will be disruptive and it could cause the equivalent of over $800 million US in impacts.  Here is the forecast impact swath …

The second question that pops in to mind after “how bad” during this season is “what about the pandemic?”  Unfortunately the things that you need to do to prepare for a tropical cyclone/hurricane are the things you don’t want to do for a pandemic.  Moving people around (evacuations) packing them together in confined spaces (shelters) are going to spread the virus and make it more likely to expose more people – and more exposure means more cases and more deaths.  The Philippines is reporting 11,350 cases (1 per 10,000 population) and 751 deaths at the moment, although many analysts are skeptical of those numbers and the government has extended their quarantine and other measures.  I expect to see a significant spike after this event – if we get good numbers.

Having fun with raw data

Here’s a plot the number of new test results per day for a number of states, normalized by population.  Notice anything odd?

Um, negative new tests in South Carolina yesterday??? and, three weeks ago, in Louisiana?  And what happened in Puerto Rico???  Are they un-testing people? Tossing out previous results? Alien Abductions(tm)?  Just plain tabulating errors?  I’m not sure exactly what is going on with this data set (the testing reports are from SC DHEC) but it shows you the perils of crunching numbers without looking at them – that would blow up any results derived from the number of people tested.  This is also why I don’t post charts every day, or plot every state/country (much less keep rolling Death Counters in the upper right hand corner of the screen updated in real time).  I wait to try to figure out what is going on first, and what the quality of the data is.  And as noted in a previous blog, it’s pretty wretched for a so-called industrialized country …

So let’s let the pandemic data simmer until next week.  Besides, there is an invest area in the West Pacific that might become a storm and impact the Philippines next week …

 

Midweek Numbers and Analysis Update (6 May 2020)

As long time readers know, one of my main frustrations is with the “News Cycle” and how there is constant pressure to come up with “new” or “breaking” news. It’s not rational or conducive to mental health.  Most events move slowly.  For Hurricanes, things change on a daily to 12 hour time scale – certainly not hourly or less – until the storm is near landfall, and even then a three to six hour update is fine.  For a pandemic, things move far more slowly.  Even daily updates are problematic – it takes time for testing results to be available, and just the course of the disease is measured in weeks as it takes 5 to 10 days for someone who is exposed to become infected (and/or infectious), another 10 days or so of illness, if hospitalized that extends to 20 days.  Hospitalization and confirmed positive numbers are a look back in time about a week – deaths, about three weeks.  Throw in the noise and inept reporting in some states, and the moving target of what is a case vs a COVID death … well, you get the picture, and it’s pretty fuzzy.

But since it has been a few days, let’s see if anything has changed – certainly the mitigation measures have, as states like Georgia have begun to lift restrictions.  But, for the reasons noted above and in a previous post, we probably won’t see any serious impact on the numbers from these measures for about another week.  So looking now is a snapshot into what was happening when the measures were expiring.

First, lets look at testing.  Despite all the talk, only a very small fraction of the population has been “tested” (noting that being tested means different things – tests for active infection have very different meanings from tests for antibodies, and all of the tests have error rates and interpretation issues as this paper demonstrates.  So how many people have been tested?  Here are the results for five states …

… only NY is above 5% of the population.  If that were a random sample that would be great, but unfortunately it isn’t.  So what fraction of those tested are positive? Continue reading