Fay Update (AL062020, Friday Morning 10 July)

Tropical Storm Fay is now off the DelMarVa Penensula as a low-end tropical storm.  It is not terribly organized, mostly a comma shaped band of rain with some gusty winds.  Here’s the satellite view  as of 7:20am … Infrared on the left, visual on the right, as always click on any image to expand.  Notice the lack of deep reds on the IR image, indicating a lack of deep convection.

The radar view confirms the picture …

Finally, here’s the damage swath. Landfall is expected tonight, but given shape of the storm that’s just a technicality.   Note the worst stays offshore; while there will be some rain and gusty winds, it is unlikely that Fay will cause much disruption aside from some scattered power outages, isolated flooding in areas that are vulnerable to flooding from a couple inches of rain, and maybe some damage from downed trees.  Given the density of infrastructure and people, economic impacts will likely be in the high tens of millions to perhaps just over $100 Million, which isn’t much for the US.

 

Tropical Storm Fay (AL062020)

NHC has started advisories on AL062020, Tropical Storm Fay.  The main threat is some localized flash flooding as it skirts the coast and moves inland over the next 36 hours.  Fay probably won’t intensify much – it’s too close to land.  Here is the forecast impact swath based on the 5pm Thursday 9 July official forecast track:

Economic impacts shouldn’t be severe, probably just over $100 Million.  Unfortunately most won’t be covered by insurance due to the named storm deductibles that are now on most homeowner and business policies.   It is unlikely that significant evacuations will be called for, which would play havoc with pandemic social distancing plans.

Possible storm forming on/near SC Coast

The system that meandered up from the Gulf of Mexico over Georgia the last couple of days (ID AL982020)  is now centered on the SC coast, although most of the convection is now well offshore.  NHC now gives it a 70% chance of becoming a depression and maybe even a tropical storm, and has even started a “Key Messages” product on it (link).  Here’s the current (2pm) Satellite view … InfraRed on the left, visual on the right, click to embiggen:

As you can see, not terribly organized. Here’s the obligatory scary 850mb wind and pressure height animation if you need a fix … click to animate.

It is forecast to hug the coast and go inland around the New York City/Long Island area, bringing gusty winds up the mid Atlantic and Northeast Coasts. Biggest problem is likely to be rain, but even that should stay mostly offshore:

Otherwise shouldn’t cause a lot of damage or disruption, so I’m afraid you’ll have to continue to worry about COVID19 and politics.

Atlantic Storms?

Over the weekend the US National Hurricane Center started tracking a system near Bermuda and last night named it Tropical Storm Edouardo.  It’s really not much of a system, at this point just a broad wind field and elongated low. Edouardo may not have a closed circulation (a key requirement to be a tropical cyclone).  But, they named it and are tracking it, so here’s the impact swath map, which will be of interest to shipping concerns and Atlanteans …

Like most of American, NHC seems to be going stir crazy, looking for something to do, and is tracking an invest area that is now on the coast of Florida, on the chances it may get enough organization to become a storm if/when it exits into the Atlantic in a few days.  They place the odds at 40% right now.  Here’s the latest track models …

Most of them keep the low inland and don’t spin up anything significant.  For those in the US Southeast, you probably won’t notice much, just a few days of thunderstorms.  Just like every other July 😛

 

Georgia 21 Day COVID-19 Death Forecasts

Nothing much going on in the tropics (couple of suspect areas off the west coast of Mexico), no recent earthquakes, so let’s take a closer look at Georgia COVID-19 fatality numbers, and how to to use them to get a look into the future.  Yes, there will be math! And an equation!!

I’m really tired of hearing people say “the spike is because of testing” or  “the fatality rates are dropping!” implying that things are not headed the wrong way. Well, the rates are dropping are a little, but that’s not helping as much as you might think.  Lets look at the ratio of people dying to new positives. That’s OK in concept, but most people doing those graphs are failing to account for how the disease progresses.  On average, it is three weeks between someone testing positive and the date they expire (and exposure to expiration time lag is probably more like 4 weeks).  If you don’t take in to account that 21 day time lag, you can reach a wrong and dangerous conclusion.  Here is what the wrong calculation (just taking the deaths and dividing by positives) ratio, and a much better calculation (taking in to account the three week delay) looks like if we plot it.  Orange is the 21 day lag ratio, and blue is the “wrong” way that is duping people …

So, yes, better treatment and more younger people catching this is probably improving survival rates.  But it’s not as dramatic as the proponents of that view suggest.  The proof, of course, is in the prediction.  Let’s go back in time to the days of yore, May 30th.  At that time the lag21 Positive Fatality Ratio was 6.111%.  It had been slowly dropping, so we compute the trend and end up with an equation that looks like this:

Forecast = (0.0611-days_since_may_30th*(delPFR – K*days_since_may_30th)*L21P
— where:
delPFR:  the change in Positive Fatality Ratio as of May 30th (2.8e-4)
K: constant to adjust for improving survival rates (calculated over 1 to 30 May, 9.2e-7)
L21P:  the number of positive tests 21 days before the date you are trying to forecast

so with this equation, we can forecast up to 21 days in the future.  It’s a very simple model (we have more sophisticated ones), but is easy to explain and only uses reported information.  How did it work, and what does it show for the next three weeks?  Here’s that plot …

Yesterday the reported deaths were 2849.  On June 11th, three weeks ago, the prediction for yesterday was 2913, only a 2.25% error.  Properly using simple tools, we can do a pretty good forecast out three weeks into the future.  It has worked reliably over the last month, so we can expect it will continue to work well unless something dramatic changes, which is pretty amazing, isn’t it?  I mean, except for the fact it is showing that because a bunch of morons aren’t taking all this seriously, and doing very simple things like wearing a mask in congested places, etc., it is very possible that almost as many people will die over the next three weeks than died over the last three months.  And we can’t do much about it because it depends on what we did in the past, and even if everybody changes their behavior today, it will take weeks for it to show up in the statistics. But aside from that, isn’t math great?  Sigh.

The coming storm?

The Senate hearings yesterday sounded pretty dire.  Epidemiologists are increasingly worried about trends in the COVID-19 stats across the south.  The positivity rates and hospitalization data are definitely headed in a bad direction.  Unfortunately, given the politicization of all this, as well as the various playing on fears people have, you’re hearing contradictory arguments like “more testing is causing higher numbers” or “now younger people are getting sick so the death rate is (or will) go down” or “masks don’t work” or even “masks are more dangerous than not wearing one.” You’re also hearing “there is so much contradictory information and what was said in March like 3 million dead didn’t happen.” As noted earlier this week, it seems that in this case misinformation is more widespread and dangerous than “fake news”.  Lets take a quick look at models and forecasts, then take a hard look at the bottom line.

First, how bad are the models?  Actually, not that bad, and they are getting quickly better.  Lets look at the country level data and a couple of the models at the national level:

Notice the difference between the black line (01 March global c19 estimate, day 30) and the Italy specific forecast on 15 March (day 45).  The early forecasts (solid black and gray lines) seem low compared to the big outbreaks, but for most countries who took aggressive action wasn’t so bad.  The forecast for Italy (green dashed line) made two weeks later is much better.  The shape different due to their actions, but the end point, and the countries with similar outbreaks like the UK and Spain who took later action, are pretty close.  That was a forecast make over two months ago, and we’ve learned a lot about the virus, and how it reacts to various public policy and personal activity decisions.  Noisy, not perfect, but useful, actionable information.

When we look at Georgia, Florida, and South Carolina, things don’t look good.  Consistently over June the cumulative number of dead on a given date is around 5% of the total who tested positive 21 days earlier.  It has very slowly declined, but when you adjust for demographics of testing it’s remarkably constant.  If we apply those trends over the next three weeks, adjust for expanding testing, etc., here is what we get. Day “zero” is June 1st, 2020, so the solid lines are what was observed in June.  The dashed lines are the forecast for the first three weeks of July assuming the 5% ratio holds …

The SC data is pretty crummy, and may be worse than it looks.  Georgia is in a slow burn.  Florida is on the verge of a doubling its death toll over the next three weeks.  It’s not good, even if the ratio is cut in half (not likely) by the changing demographics of younger people being the primary source of new positives.

Opinion and recommendation: I have mixed feelings about local city government mandating actions like masks.  But this isn’t really the place to argue about the legal basis for these orders, or how in a community with complex overlapping jurisdictions like, for instance, Chatham County Georgia.   It’s confusing and far less effective when one municipality mandates it and the others don’t (that’s a major reason why the shelter in place orders and reopening were ineffective, they weren’t uniform). Forget what your mayor says one way or the other.  Forget the pedantic, technical nit-picking arguments over how effective masks are. The science is evolving, and there is a lot of seemingly contradictory information out there that, in context, does actually make sense.  The bottom line is that even if only 20% effective at reducing transmission (and the vast majority of studies indicate that cloth masks are at least that good, and the layered stuff is even better than that), it has a good chance of dropping the reproduction rate below one, which will get the pandemic under control.  Yes, N95 type masks do more to protect you than others.  Yes, cloth or procedure/surgical masks mostly protects others.  But the procedure masks do provide you with some protection – just as the N95’s do provide some protection for others.  Yes, you can conceivably catch the virus from improperly taking off your mask – but that risk is a small fraction of your risk from catching it without one (note this is different in a clinical environment, where self-contamination is a bigger concern).  So please just wear the mask when you go out in public to confined places, continue with good hygiene and social distancing. Learn how to properly put it on, wear it, and take it off.  We know how to beat this thing, if people would just be sensible.  But … remember actions today take three or four weeks to show up in the mortality stats.  July is likely to be a rough month, with lots of fear, angst, and finger pointing.  Sigh.

About masks …

Misinformation kills.  The problem is that most misinformation is on some level true.  That seems like a contradiction in  terms, but that’s how it is.  Take the arguments over masks.  I have seen various “experts” talking about how cloth or surgical procedure masks don’t filter out virus particles, so they are worthless for the SARS-COV-2 virus that causes COVID-19.  The first half of that statement is true.  The second half is dangerous BS.  It is vital to keep in mind that the virus is being spread in small droplets when you talk, sneeze, cough, even just breath (ewww).  “Small” is relative; those “small” droplets are  millions of times larger than the virus. Stop the droplets, stop the virus. The TL;DR is that if you have COVID-19 and wear a normal procedure mask, you reduce your chances of spreading it to someone else by something on the order of 80 or 90%.  If you wear a mask it “only” reduces your chance of getting it by 50% or so.  If both are wearing?  Estimates are over 95% or so effectiveness – maybe 99%+.

Here is a short take on the subject from a Stanford Medicine scientistHere is a more comprehensive scientific literature review in PNAS.    So just wear a (insert colorful language) mask in public places – cloth is fine.  Of course, in a clinical setting, you should be using an N95 mask and face/eye protection.  But Publix isn’t a clinical setting.

There is a French saying:  if you put a spoonful of wine in a barrel full of sewerage, you get sewerage.  If you put a spoonful of sewerage in a barrel full of fine wine, you get sewerage.  So it is with information … keep the sewerage out of your wine.

Speaking of sewerage, here are some updated charts.  The trends are clear, Arizona, South Carolina, and Florida are in full blown rapid community spread. Texas is on the verge of that.  Georgia is trending in a bad way.  Expect a spike in hospitalizations in a week or so, and deaths in a couple of weeks …

This first graph is the rate of new positives.  It’s noisy, when people show this data they often smooth it – I don’t because the noise is information, and shows you why you can’t get excited about day to day moves until the trend is clear.  This is the rate of new positives per capita (per 1000 people).  That way you can compare states; 100 people in Arizona is about three times worse than 100 people in New York (pop 7 million vs pop 19 million).

This next graph is the cumulative positives. This is per 10,000 people. The shape of this curve tells us where we are in the pandemic.  These are called logistic curves – they tend to have a characteristic “S” shape.  When they start to bend sharply upward (like AZ, SC, FL) it’s bad news … in short, the previous graph is showing how fast the graph below is changing.

Misleading numbers, more dust in the wind

I’d rather be writing about hurricanes (or the lack thereof), or the dust (some of which is now over the southeast), but … at the risk of annoying folks, I think it’s worth talking about Florida, testing, and why more testing isn’t the reason there are more positives (even though that’s “true”).  But read on and there will be an embiggenable dust image and notes at the end.

There is a major misconception going around that the reason for the increase in the number of positives is that there is more testing.  This is one of those things that is true, but misleading.  So it’s worth a closer look, using Florida as an example because their numbers are now trending into “New York in April” territory if something doesn’t change.  Two weeks ago (June 12th), Florida had tested 1,335,899 people.  As of yesterday, they have tested 1,768,885 people – a 32 percent increase.  But the number of positives increased from 70,971 to 122,960 – a 75% increase!  Put another way, the positive test ratio (a key metric in epidemiology) consistently increased from 5.31% to 6.95%.  In simple terms, the virus is still expanding faster than the testing.

Let’s contrast that with New York.  Over the same two weeks, New York expanded testing by almost the same amount: 31 percent.  But … the number of positives only increased by 2%, and the positive test ratio consistently dropped day by day from 13.63% to 10.63%.  In other words, as New York tests more people, a fewer percentage of those people are testing positive.  So the testing is “getting ahead” of the expansion of the virus.

What about Georgia?  The positive test ratio was at 9.42% on June 12, had dropped to 9.18% on the 18th, but is now back up to 9.47%.  Based on other data, that’s about what we expect, that 9 or 10% of the general population should be testing positive for this thing.

Some people are latching on to the fact that the hospitalization and fatality rates are dropping, and attributing that to the younger ages of people testing positive.   The ratio of dead to positive does seem to have plummeted from 4.18% to 2.82% in Florida.   But that may be misleading because of the 20 to 30 day lag between testing and mortality.  if we use a 20 day lag, the number has been pretty constant over the last two weeks, dropping much more slowly from 5.92% to 5.52%.  So let’s not get excited about that just yet …

OK, enough virus – let’s talk dust.  While this is a larger than normal plume, it’s not terribly unusual.  I think folks are more attuned to troubling signs and portents these days (you know it’s weird when people start calling grasshoppers locusts).  While people with respiratory problems might have some problems (especially in the Caribbean, although in the SEUS as well but since you are wearing your mask 😛 …), it shouldn’t be much more than a light show for us at sunrise/sunset.  Their are two reasons why this plume is more substantial than in most years.  First, it has been a bit drier and windier at the surface in Africa, which picks up more dust.  Second, the middle and upper level winds are more consistent, and stronger, blowing across the Atlantic.

Here’s the promised dust image as of 8:40am this morning … as always, you can click to make it larger (embiggen, for you scholarly types):

Trends or Noise? A brief review of how to look at data …

ZOMG! TODAY IS ANOTHER RECORD DAILY INCREASE IN COVID CASES AND THE SECOND RECORD IN A ROW!  Or, it’s a leveling off of the increase.  Or … well, what is it?  It sort of depends on how you look at the data.  I’ve spilled lots of electrons on how the “case” data is really problematic, but that’s what people are using for decision making, and it’s closer to what the virus is doing in “real time” (only about two weeks behind reality, vs. the mortality rate, which lags about a month).  So let’s dig in … sorry, no equations.  

First, the raw numbers and “death counters” you are seeing everywhere can be very misleading.  The spread of a virus through a population depends on a lot of factors, but one of the biggest is population density.   You HAVE to normalize these numbers for population for this to make sense.  100 people in New York City is a blip.  100 people in rural Georgia is a catastrophe. Likewise, two months ago when there were over 20,000 new cases a day, they were mostly in New York and New Jersey; today, they are scattered across a dozen states.  So here is what we are seeing in representative states for the rate of new positive tests (people are calling these a “case” but that’s a misnomer).  This graph shows new positives per 10,000 of population.  Yes, there are some negative numbers and zeros due to reporting issues … (SC and LA are really crummy data sets).  Click to embiggen.

So in relative terms, most states are seeing a spread of the virus on the order of 1/3 the rate of what was seen in NY/NJ at its peak.  Note that’s not a good thing … but, again, 30k people infected in NY/NJ (and most of that in the Metro NYC area, population of about 20 million) vs 30k people across the entire US (population 321 million) is a very different thing. But … we are seeing some states (TX, GA, SC) doing much worse, not only moving in the wrong direction, but into the realm where you have to start worrying about the health care system getting stressed.

Let’s look now at the country as a whole:

Notice how flat this curve looks compared to the state level.  That said, there is a disturbing trend over the last 10 days, however, it’s not really the dramatic upward spike that seems to be the perception in some circles. And as I have often said, single day trends, even week trends, can be misleading.  Is today a plateau, followed by a downward trend, or will the rate increase?  My guess is it will wobble a bit higher then trend back down, and the “stable” point nationally over the summer will be around 1 per 10,000 per day in a “somewhat open, but with most people being sensible” paradigm. That’s not as good as the previous somewhat “stable” point of around ~0.75 per 10k for the last two months, but not horrific.  The trick in looking at this data – and creating policy to address the pandemic – is figuring out the trends, with the knowledge that with new positives you are two weeks behind the curve.  The data point for tomorrow depends on what people did 7-14 days ago.

In short: absolute numbers may be dramatic, but it’s the trends in rate per capita that matters.  Speaking of which, absolute hospitalization and death numbers, while vital to gauge the stress on the health care system (which is getting worse in many states, especially where the positive rate is moving above 1.5 per 10,000), aren’t telling you how deadly or dangerous the virus is.  As noted yesterday, the rate of people who are positive that end up in the hospital, and the rate of people who are dying from the virus, is actually decreasing slightly.  This is likely because more people in the 20-50 year old range are the bulk of the new positives.  So while the news coverage has shifted this week, what we are seeing isn’t a dramatic change in the virus.

The trend over the last 10 days is of course being blamed on reopening, and that is a big factor, but more importantly these trends are a direct result of people relaxing basic precautions – especially not wearing masks in public places and gathering inappropriately.  The reopening didn’t have to cause this increase.  In my view it is an urgent call to take more precautions on an individual level. (Update: and, an hour or so after I posted this, guess who said the same thing, but obviously a bit better …)  I don’t think it’s a call to close things down again; our economy is very fragile (likely in an ongoing collapse with the worst yet to come) and just can’t handle that.

Finally, I’m not trying to minimize how bad this is, or get into the partisan poo flinging over it.  As for those saying that the positives are related to testing; the argument that more testing is the reason we are seeing more positives  is “not even wrong.”  But that said, we need to have some context about this pandemic, and chart a careful path between over-reaction and under-reaction.  Sadly, all we seem to be doing is a random weave between the extremes … and that has severely damaged our economy, and maybe even our society as people are increasingly frustrated and angry as pre-existing fault lines are exposed and aggravated.

The Second Phase of the First Wave (COVID Update 25 June)

The last day or so have seen disturbing headlines about the “surge” in COVID-19 cases, “record new US cases,” and the NYT this morning has the calm and rational headline “How the Virus Won.”  So how bad is this?  Like so many aspects of this pandemic, the answer is “it depends on how you look at it.”  But it’s not good, that’s for sure.  Let’s look at the data which, as I have ranted, are embarrassingly  bad for a so called developed country, and see what we can see.

First, let’s look at the trends in people testing positive for the active virus.  With the caveat the data is still a mess (antibody and active virus tests are still not clearly split in some jurisdictions, and testing positive isn’t necessarily a “case” in the classical definition due the high asymptomatic rate with this thing, etc), it’s not headed in a good direction in many states:

I’m deliberately not smoothing this because I think smoothing makes you think the data is better than it is, and masks how this virus seems to work (super-spreaders, big jumps up and down in the spread rate).  But you can clearly see two groups of states: the initial outbreak, rapid spread states of NY, NJ, Louisiana (mainly New Orleans), and the “slow burn” states (everybody else).   In the “slow burn” states like Georgia and South Carolina we are seeing a really bad trend over the last 10 days. We aren’t seeing New York/NJ/LA kinds of spread – largely due to the much lower population densities, and the fact that 30-40 percent of the population are still taking mitigation measures.  But let’s pull out the overall national trend …

It’s disturbing, but not panic-worthy yet.  As can be seen, other than the initial run-up in NY/NJ, there have been previous quick jumps similar to this. I *think* it’s real, and *potentially* the start of another run-up, *definitely* a wake up call, but the data is so bad I’m not willing to jump on that bandwagon that this is the start of another NY style out-of-control phase 2 of the outbreak just yet.  If it persists through the weekend then we can start to freak out.  Here’s the overall confirmed positive (viral only, with some contamination):

In the “good” news category, the hospitalization rate (how many end up in the hospital) and mortality rate (how many who catch the virus die) are pretty stable, and the trend is not bad.  Note that in absolute numbers, yes, the numbers are headed higher – Georgia reported 1124 COVID-19 patients in-hospital yesterday – the highest since May 16 (and it is likely higher given the delayed/bad reporting).  But the probability of the virus being fatal, if you test positive and end up in the hospital, is actually down slightly.  This is likely because many of the most vulnerable have already fallen, or are being protected somewhat.  So the virus isn’t getting worse, it’s just spreading.

Cautionary note: I’m seeing folks looking at ZIP code level and city level data, such as in the Savannah GA area.  Don’t waste your time.  At that level it is so noisy and contaminated it is worthless.  County level you can sort of maybe draw some conclusions from, but with a lot of caveats.

Internationally, the story is mixed.  Much of Europe has leveled off. The big outbreaks, Spain, Italy, UK, are leveling off (albeit beware data issues in Spain).  Sweden continues a jumpy climb but the rest of the Nordic countries have leveled off and have things under control. Russia’s trajectory is still in the “slow burn” category nationally; Moscow Oblast, which is driving their numbers, seems to be leveling off, so in that sense is similar to the US with a couple of big cities driving the overall numbers. Data issues really cloud the picture in most of the rest of the world.  On the surface, the US is in the lower tier of countries when normalized by population, but as seen above, the virus is in the details.

So what do you do?  The advice hasn’t changed much.  Mask up in public, social distancing, good hygiene.  Avoid distant travel if at all possible, avoid groups, stay in your social “bubble.”

Administrative note: I’ve set up a dedicated Twitter feed for Enki Research if you prefer to get your doom that way: @EnkiResearch