2020-05-09

Coronavirus days: regionalism, modelling, hypocrisy, global warming

MVIMG_20200508_160649 Let's start with Neil "Bonking" Ferguson, who said: "I accept I made an error of judgement and took the wrong course of action. I acted in the belief that I was immune, having tested positive for coronavirus and completely isolated myself for almost two weeks after developing symptoms. I deeply regret any undermining of the clear messages around the continued need for social distancing." This is interesting, in the light of claims to be "following the science". It would appear that his scientific advice is that, if you've had the virus you're immune so contact is permissible. As far as I can tell he believes his error was "undermining the message" rather than his scientific judgement. This then recalls... The Climate Change Hypocrisy Of Jet-Setting Academics? No, not that one, but the linked Climate chickenhawks.

But perhaps his scientific judgement was correct. Of course in the fervid atmosphere of the UK that is no proof against charges of hypocrisy, as should have been bleedin' obvious to him. Which leads me on to regionalism. Or, that the current one-size-fits-all policy is, errm, how about "non-optimal from a theoretical point of view". The rates of infection in the country vary wildly by region, with the general pattern being that outside the plague-pits of large cities things are fairly OK. And using hand-waving density arguments, that would likely be true with less lockdown. Unfortunately the centralised response we have appears to be unable to think about this; R4 this morning were desperately worried that even the smallest divergence between England and Wales might be confusing. Perhaps it might confuse their tiny minds but I think people could cope. The USA gets to experience this more directly I think, since the States are responsible for themselves, and this is good. Closer to home, we're going to see changes in Europe as an example - for either good or ill - to us.

One Adam Kucharski sticks up for NBF Twatting There seem to have been some misconceptions about how COVID science is done and how it helps inform decision making. As we clarify here, it's a large collaborative effort... and linking to their doc (arch). Which includes inter alia In early March 2020, the emerging consensus amongst scientists involved in this country-wide consultation was that SARS-CoV-2 was circulating widely in the UK, it was capable of causing substantial hospitalisations and fatalities, and that in the absence of drastic social distancing measures, the healthcare system would rapidly become overwhelmed in the same way that it had been in Northern Italy at the time. Although new studies and data have since emerged, this consensus has not changed. But there are problems there. Firstly the idea that little has changed revises away rather large changes in doubling time, as JA noted. And secondly, if the govt is "following the science" and the science in early March was such, why did we lock down so late? The answer of course is that rather important things did change, most obviously their belated revision of the doubling time.

Incidentally, SM points to the list of meetings which contains lots of interesting but frustrating minutes; it would be good to pore over those some time - but even better for someone else to do it and synthesise their timeline. Having browsed a couple I found myself frustrated because there are raw statements not backed by any references.

Moving now onto models, consider the Imperial model. Obviously at a superficial level, because I know little about it, but I did find We modified an individual-based simulation model developed to support pandemic influenza planning to explore scenarios for COVID-19 in GB. The basic structure of the model remains as previously published. In brief, individuals reside in areas defined by high-resolution population density data. Contacts with other individuals in the population are made within the household, at school, in the workplace and in the wider community. Census data were used to define the age and household distribution size... This model has been a bit crap overall, because as JA has so elegantly demonstrated, they fucked up the calibration by not doing any. Indeed this seems almost a textbook illustration of how useless it is to have an elaborate model capable of simulating fine-grained detail when important controlling parameters aren't know. As a country-aggregated simulation it appears almost comically inappropriate. And yet for what I want them to do - modelling regionally-varying degrees of lockdown / isolation - it could be quite appropriate; but the little buggers aren't doing that.

Global warming


Richard-the-Betts Twit “Like in the Covid pandemic, timing [of climate action] is critical to prevent devastation. If you wait until you already have a serious problem, then it is too late. Unlike with corona, sea-level rise cannot be stopped for many centuries", riffing on Sea levels could rise more than a metre by 2100, experts say. I disagreed, on the grounds that there's no exponential growth of temperature. In a thrilling continuation, RAB hit back with But there is a very long lag in sea level response. If we wait until we've seen (say) a 50cm rise before doing anything, we'll be committing future generations to substantially more, but so far he has no response to Well that's part of the point: the lag really is very long indeed. There is some limit beyond which it is not worth planning: Should we care about the world after 2100?

Bees


Ah, my picture: a swarm from one of my hives. They flew off not very far and collected themselves on the muddy bank of a local brook. We collected them into a large cardboard box and tookthem away; Mac reported that they swarmed again this morning, but he seems to have re-collected them again.

Another view


From Kal.
   

27 comments:

  1. I disagreed, on the grounds that there's no exponential growth of temperature.

    But Betts example was sea level rise which is almost certainly going to be much faster than linear if not exponential.

    ReplyDelete
  2. That didn't occur to me... but if I take the data for 2100..2500 from https://www.ipcc.ch/site/assets/uploads/2018/02/WG1AR5_Chapter13_FINAL.pdf table 13.8 I get a good fit to linear (for the upper and lower bounds) and a clearly poor fit to exponential.

    However, the actual mathematical form of the fit isn't desperately interesting (any sufficiently slow exponential is well approximated by a quadratic...) what's interesting is how fast it is growing; and SLR isn't growing "quickly" in any Coronavirus sense.

    ReplyDelete
  3. Maybe sea level rise isn't the best way to compare the response to Covid pandemic and the response to AGW, but I think Richard Betts' basic point is reasonable. Given that the impact of AGW is likely non-linear, there are significant implications to delaying when we start to reduce emissions, just as there are significant implications to delaying when we went into lockdown. Or, maybe, in both cases each unit of delay doesn't have the same impact.

    ReplyDelete
  4. Table 13.8 is based on ice sheet models which, it says in the caption, are considered to be crappy for modelling the dynamics of the Antarctic ice sheet. You're basically reading out the results from the uncalibrated covid models and saying everything's fine.

    There are two exponential factors involved in sea level rise. One arises from largely linear factors (thermal expansion, glaciers, ice sheet SMB) in concert with inertia, meaning that there is committed SLR "in the pipeline", and as forcing keeps going up fairly linearly you get a linear rise in growth rate, maybe up to 15mm/yr by 2100. Whereas if you apply strong mitigation now to quickly stabilise forcing we'll see no change in SLR rate over the next Century (see Figure 13.11b).

    The other potexponential comes from dynamic losses of a somewhat unstable West Antarctic Ice Sheet (also increasingly suggestions that the East Antarctic may not be as stable as previously advertised). The more we push now the greater risk we expose ourselves to later.

    I doubt Richard Betts was suggesting there is an exact 1-to-1 correspondence, but just that decisions now make a big difference, and the longer you wait the less difference you'll make.

    ReplyDelete
  5. > Table 13.8

    Fine; in that case, pull some more reliable numbers out of the report instead :-).

    > Given that the impact of AGW is likely non-linear

    That's giving a lot. In fact, temperature response is nearly linear in exp CO2, which is to say, up to about now, nearly linear in time. Ah, but you mean "impacts" are non-linear in temperature. This is a fair point. But I'm still very doubtful that the analogy is useful; in Covid, a week was a long time; in GW, arguably a decade wasn't.

    ReplyDelete
  6. but you mean "impacts" are non-linear in temperature.

    More than "mean", actually "said" :-)

    I probably agree that the difference between emission reductions starting now, and emission reductions starting in 10 years time won't be that great. However, I do think that the impact of each additional decade increases non-linearly (i.e., the impact of a delay of a decade is greater in the future than it is now).

    ReplyDelete
  7. Given that the impacts to date of a warming planet are best described as 'slight' (longer allergy season, premature alteration of migration patterns, undeniable shrinking of Arctic ice, etc.) it almost becomes an assumption of those most concerned about future impacts that they will be non-linear. Certainly the IPCC's description of future impacts would not be described as non-linear.

    It seems to this non-scientist that the non-linearity comes from the Deux ex machina-like interventions of things like WAIS. But the destabilization of the WAIS certainly preceded our contributions to global warming and is in fact due to mechanical stresses that were described in the 1920s without any reference to a changing climate.

    Clearly there is something in the discussion about impacts that escapes me. Too many credentialed and clearly intelligent scientists are concerned about it to suggest otherwise. But just as clearly they have failed to make the case to me and apparently enough others to make a political difference.

    ReplyDelete
  8. > the destabilization of the WAIS certainly preceded our contributions to global warming and is in fact due to mechanical stresses that were described in the 1920s without any reference to a changing climate

    That sounds unlikely. Got a cite?

    ReplyDelete
  9. Tom,
    Clearly there is something in the discussion about impacts that escapes me. Too many credentialed and clearly intelligent scientists are concerned about it to suggest otherwise. But just as clearly they have failed to make the case to me and apparently enough others to make a political difference.

    I think the simple argument would be that there is some level of global warming that would be utterly catastrophic, and that this level is probably less than 10C, but bigger than 4C. The exact number is probably not that important. There's also a level of warming (from pre-industrial) that would lead to modest impacts; say 0.5C - 1C. Given this, the impact of each extra increment of warming would then be non-linear. In other words, if the impact of 0.5C is negligible, and the impact of ~10C is catastrophic, then the impact of each extra 0.5C has to (on average) increase with increasing warming. To be clear, the numbers I've used are illustrative, not exact.

    ReplyDelete
  10. Well, The Comparison of Pleistocene and Present Ice Sheets is from 1892, but it's paywalled.

    https://pubs.geoscienceworld.org/gsa/gsabulletin/article-abstract/4/1/191/3194/Comparison-of-Pleistocene-and-Present-Ice-Sheets?redirectedFrom=fulltext

    Similarly, 'Antarctic and Some of its Problems' from 1914 is captured by JSTOR and they want money.

    Anyhow, there are 1,640 articles on the Antarctic ice sheets on Google Scholar prior to 1930. I can't find the ones that caught my attention a decade ago, but I hope you'll accept my solemn oath that I did read academic papers speculating on the destablisation of the WAIS due to mechanical stresses way back when...

    ReplyDelete
  11. Fine; in that case, pull some more reliable numbers out of the report instead :-).

    The report doesn't give much in the way of concrete numbers in that respect. Figure 13.11 gives the medium-high confidence results then they non-noncommittally state that there could up to another several tenths of a metre due to dynamic losses from unstable ice sheets this century.

    That sounds unlikely. Got a cite?

    Unlikely is generous. Based on IPCC reports glaciologists have been very uncertain about WAIS instability until fairly recently, so it would be very odd if a study from the 1920s provided absolute certainty on the matter.

    Also, I'm not sure talking about "destablisation" like that makes sense in terms of what is meant by stability here - it's a function of how the glacier system dynamics work in the conditions around the edges of West Antarctica.

    ReplyDelete
  12. could up to another several tenths of a metre due to dynamic losses from unstable ice sheets this century.

    Forgot to add that this would mean SLR at about 10x current rate towards the end of the century.

    ReplyDelete
  13. Also, even a linear rate of SLR results in an exponential growth in coastal flooding risk. Doubling time gets shorter the higher the emissions scenario.

    ReplyDelete
  14. > 'Antarctic and Some of its Problems' from 1914

    Oh hold on. The *concept* of the WAIS being unstable could have quite an extended life (but bear in mind that people knew bugger all about it's bathymetry at that point). But that was only at the vague ideas stage. And anyway your assertion as I understand it was that "destabilization of the WAIS certainly preceded our contributions to global warming" -i.e., the actual destabalisation, not people speculating about it - which I don't think you've justified, and which I don't believe.

    > SLR at about 10x current rate towards the end of the century

    It is still hard to get more than a meter this century and... I don't think that will be enough, TBH. In the sense that it won't weigh heavily enough in people's minds to matter. At that level, temperature rise becomes more interesting.

    ReplyDelete
  15. Sorry about any confusion, but no--it was the conversation about destabilization occurring back in the day, not any real destabilization.

    ReplyDelete
  16. "It is still hard to get more than a meter this century ..."

    Assuming the WAIS melts in place, yes.

    Assuming the answer doesn't convince anyone.

    So will it??

    https://blogs.egu.eu/divisions/cr/2016/06/22/marine-ice-sheet-instability-for-dummies-2/

    ReplyDelete
  17. I don't think that will be enough, TBH. In the sense that it won't weigh heavily enough in people's minds to matter.

    Isn't that the point Richard Betts is making? Three months ago viruses didn't weigh heavily enough on people's minds to matter, yet here we are now.

    Maybe if someone lived in Birmingham they might reasonably think it doesn't affect them directly, but how about Venice? Though even for inland folk global trade and the global economy are highly dependent on ports which will become more vulnerable to extreme coastal flooding events.

    As I've pointed out, the issue is not so much the average rate of sea level rise but the rise in concert with normal variability which results in exponentially increasing exceedance of the infrastructure thresholds we have in place. If you want IPCC numbers see Figure 13.25, shows increased frequency in coastal flooding events around the world given a 0.5m rise. In quite a lot of places you're looking at more than 1000x increase in coastal flooding events, just from 0.5m.

    ReplyDelete
  18. PaulS, and yes you point out one of the very plausible and easily predicted of the global warming we are likely to experience. Vulnerability to storm surge, etc.

    It is predicted to make X miles (forget the exact number) of coastline unusable for the purposes we now employ it.

    It's a very small percentage, though, less than one. The EPA predicted back in 1992 that 1 meter of sea level rise would render something around 1% of coastline unusable. They also costed out adaptation at a very low cost--sea walls and moving roads and buildings.

    ReplyDelete
  19. Tom,

    The EPA predicted back in 1992 that 1 meter of sea level rise would render something around 1% of coastline unusable.

    That sounds like a statistic about general sea level rise inundation rather than increased vulnerability to extreme sea level events. In any case a percentage of total coastline is not meaningful since the vast majority of the world's coastlines are empty of people and their stuff anyway. And you'll find there's a very strong correlation between vulnerability and where people and infrastructure actually are along the coast. Because having easy access to the sea is useful.

    They also costed out adaptation at a very low cost--sea walls and moving roads and buildings.

    Well, that's part of the point. If it's so easy and cheap why isn't all that stuff happening now? And why didn't it happen before Storm Sandy hit, for example? Are we waiting for further damage before we actually do anything? It's really not enough to just say "Well, we could do this so problem solved and let's never speak of this again."

    ReplyDelete
  20. Hi Paul

    Umm, perhaps because we tend not to fix the roof when it's not raining?

    ReplyDelete
  21. Tom,

    When it's not raining is the perfect time to fix the roof. What's the point in quoting low costs of taking preemptive adaptation measures if you don't believe those are going to happen?

    Interesting little case study is the Dawlish rail collapse. In 2014 a storm surge overwhelmed the existing century-old sea wall and swept away the ground beneath a 50 metre long section of railway. This is/was on the mainline for both freight and passengers between London and Devon/Cornwall, and the loss of the line for several weeks consequently was estimated to cause a loss of over £1 billion. Initial repairs to restore the track to working order cost £40m. The site has subsequently seen increased frequency of flooding at extreme high tides, resulting in occasional line closures and probably further losses in the hundreds of millions.

    Five years later in 2019 work began on a sea wall upgrade designed to withstand sea level rise for the next hundred years. Still not yet complete but supposed to be in the next few months. £80m was budgeted though supposedly the new sea wall itself may cost as little as £30m.

    So, yeah, a purely adaptationary response to sea level rise with regards this 50 metre stretch of coastline could have theoretically cost £30m but in reality cost about £1.5billion because we waited until after the rain fell to fix the roof.

    ---------------------

    Another interesting case study is the MOSE project in Venice. Originally proposed in the 1980s, work started in 2003 and currently expected to be completed in 2022. Originally supposed to cost about 1.5 billion euros, now estimated at 6 billion.

    It's been extremely controversial due to some corruption scandals, plus apparently reasonable misgivings from some locals about potential effects on the ecology of the lagoon. There's even a conspiracy theory among some locals that the system itself is the cause of increased flooding events.

    This could be an example of preemptive adaptation, but it's taken so long that events it was supposed to prevent have already happened during construction, most notably the extreme flooding late last year. That single flooding event may have cost Venice as much as the MOSE project itself in damages and lost tourism. There have been lots of cries of "Where is MOSE?" after the event but not so much beforehand.

    But even when it's up and running there are some known issues. Because of aforementioned ecological considerations the operational parameters will be such that the system will only be triggered at a tidal threshold to protect against widespread flooding. Flooding of low-lying areas such as St Marks Square will continue unchecked, and will continue to increase exponentially with sea level rise, while also increasing in average severity.

    Events which do trigger the system will likewise increase exponentially. As a mechanical system this means an exponential increase in wear and tear and maintenance costs. Given spiraling costs and the fact that it wouldn't have "rained" for a long time it would be very easy for them to forego the effort to fix that roof properly, which could have catastrophic consequences.

    ReplyDelete
  22. > the loss of the line for several weeks consequently was estimated to cause a loss of over £1 billion

    That sounds implausibly high, and isn't obviously in the article you ref. The BBC quotes a report claiming 1.2B (https://www.bbc.co.uk/news/uk-england-devon-31140192)... or more precisely, "The forum, which is behind the report, said the economic impact, which includes the tourist and fishing industries, is estimated to be anything from £60m to £1.2bn". That's rather different. Actually it is even vaguer than that: "Estimates of the economic impact of the mainline rail closure at Dawlish, for example, vary significantly from 1 - 20 million
    pounds per day or £60 million - £1.2 billion over the course of the two month closure" (https://devoncc.sharepoint.com/sites/PublicDocs/Environment/_layouts/15/guestaccess.aspx?guestaccesstoken=aH%2F6l0ldogPV0ifEULLEFPAs9KqKUqxpn7EMrpHDgz8%3D&docid=0f0be37662cbf4c84992e0415a7b5991a) which in turn comes from "These estimates are based on an extrapolation of the widely reported £20 million per day figure (see for example http://www.westernmorningnews.co.uk/Severed-rail-link-cost-region-20m-day-Devon/story20575620-detail/story.html". Which isn't available.

    So I don't think your figure can be considered reliable.

    ReplyDelete
  23. That sounds implausibly high, and isn't obviously in the article you ref.

    It's a widely cited figure, often also expressed as £20m per day, attributed to regional Chambers of Commerce. We're talking about basically two whole counties being cut off from the rest of the country by rail for two months, sounds plausible to me.

    The core source of the £1.2 billion seems to be a survey of businesses centered around Plymouth, suggesting a probable £1m a day cost just to businesses in Plymouth. Extrapolating to businesses around the whole affected region apparently led to a figure of £20 million per day. It's not clear whether the £20m figure comes solely from these early survey results or an expanded survey of the region later.

    I can't find any concrete sources at all suggesting anything lower than that, so not sure where the £1-20m range comes from. Maybe confusing the estimate for Plymouth alone with the whole region?

    Of course it's very difficult to estimate precisely, but I can't see any justification for considering figures below several hundred million and it could be higher than £1.2 billion. That figure doesn't appear to fully include potential longer term impacts outside of the closure period - losing contracts, goodwill impairment etc.

    ReplyDelete
  24. I don't know how far it is worth chasing this. "a probable £1m a day cost just to businesses in Plymouth" is dubious; a fuller quote is "Over 200 Plymouth based businesses participated in the survey, with results indicating that the majority are facing a minimum daily cost of between £100 and £1000, with others sighting [sic] costs of in excess of £5,000 for each day that the Dawlish line is closed. Aggregating this figure for the 6,000 VAT/PAYE registered businesses in Plymouth, it is possible to assume a cost to the city’s economy of at least £600,000 each day, which could almost certainly rise to over £1m per day as the continued lack of a working rail line takes effect".

    Presumably, those most likely to reply to the survey would be those most affected, so aggregating up to 6k business seems dodgy to me. Also, as time went by people would likely find workaronuds.

    However, if you point is that preventative maintenance is a good idea, I'm not disagreeing.

    ReplyDelete
  25. I'm not saying that ignoring the roof in fine weather is an optimal strategy--just that it is what humans tend to do. And of course when it rains it's too wet to fix the roof. Just sayin'.

    ReplyDelete
  26. PaulS,

    "Also, even a linear rate of SLR results in an exponential growth in coastal flooding risk. Doubling time gets shorter the higher the emissions scenario."

    The make the exact same basic error I've now seen countless times. Oh and it is in Scientific Reports, so it is obviously junk.

    No waves and no storms and no storm surge. So that all that paper is good for is nuisance flooding due to tides. Things that happen annually (say two days a year increase to four days a year then to eight days a year, now that is the definition of doubling time and the numbers I used are just a for instance type scenario).

    No one in the coastal arena designs anything based on tide gage data alone. NO ONE!!!

    What that paper clearly does not understand is that a current 50-year event is not due to just the changing tides. Those higher water levels are almost certainly due to actual storm events. One needs to do joint PDF's obtained from actual hindcast models. No one should simply lift the FOO graph by a uniform amount. EVER!

    ReplyDelete
  27. From that POS paper ...

    "In the current study, we focus our analysis on extreme water-level events (due to tides and storm surge but not waves) at the present-day 50-year return period, since most coastal engineering works in the U.S. are designed for return periods of 50 years or less30."

    The CEM is OLD!

    What is actually done is a 100-year event using a 50-year C/B analysis (the improvement must pay or itself in a 50-year time frame but it is designed for a 100-year storm, that is currently standard practice).

    There are nowhere near enough tide gages necessary to do a proper design, even if the design is right at the tide gage, hindcasts are always performed.


    Now, if you wish to speak about dumping sand on the beach then Homey don't play that as Homey is not, has never been, and never will be, a SAND ENGINEER (call that one an in-their-face pejorative because Homey absolutely HATES sand engineers).

    ReplyDelete