The UK weekly infection rate has been falling since the 18th November and is now about the same as that in Germany (150/100,000). Note that infection rates had mostly stabilised already by early November.
Weekly Infection rates comparison (derived from ECDC data).
The fall in rates for France has been even more dramatic from a peak of 570/100,000 on 8th November now down to below that of Germany’s in just 4 weeks. Italy too is well past the second peak and falling fast. Germany however is now suffering worse during the second wave than it did during the first wave.
Mass vaccinations will naturally bring infections rates down as the pool of susceptible people reduce. If we could perhaps vaccinate >20 million people by Easter then the epidemic would essentially be over and normal life can probably resume. It can only end with “herd immunity”.
The global average temperature anomaly for October was 0.706C, which is a fall of 0.15C from September. This reduces the annual average so far to 0.9C leaving 2020 still slightly higher than 2016. However the uncorrected data (without pair-wise homogenisation) leave 2020 just below 2016. My calculation of the global temperature anomaly is based on GHCN V4 and HadSST3 using a 3D spherical triangulation method and a baseline of 1961-1990.
Global average temperatures (anomalies) where 2020 is averaged over 10 months. The green points are the uncorrected temperature data
The monthly data below shows a large drop in temperatures for October
Monthly average temperatures V4U is the uncorrected GHCN V4 data and V4C the corrected version using pair-wise homogenisation.
The spatial distribution below shows lower than average temperatures across North America., Central Asia and Western Europe. Blue colours show temperatures below the 1961-1990 average for October.
Spatial temperatures in the Northern Hemisphere
There were also cooler ocean temperatures in the Southern Hemisphere.
I have noticed an interesting effect of the pair-wise homogenisation process. Recent months seems to show a large divergence between the corrected and uncorrected GHCN V4 results. However this difference slowly decays with time so that past differences reduce. This is because the corrected data from previous months and years also slowly change as the homogenisation algorithm is rerun each month. This seems to produce a self correcting process tending to reduce strong discrepancies. over time.
The news this morning is that the government is due to announce a nationwide lockdown on Monday because deaths are even worse that foreseen in SAGE’s worst case scenario which was leaked 2 days ago. The Spectator on Thursday published the SAGE modelling group’s Reasonable Worst Case Scenario (RCWS). This also explains why Boris Johnson had been so cautious about opening up the economy. France, Germany and Italy have all announced new lockdown measures. Is all this actually justified?
Firstly the UK is actually doing better than the rest of Europe and cases are growing more slowly. They appear also to be levelling off under the current regional “tier” system.
7-day infections per 100,000 population in different European Countries.
Just as the current measures seem to be working, Boris is under new pressure from SAGE for a new national lockdown because deaths currently exceed that foreseen by their modellers in their RWCS at the end of October. Their scenario shows the second wave of Covid-19 starting this autumn with a long slow peak lasting until March causing up to 85,000 deaths and a surge in ICU beds . The first version of the RWCS report was published in July. No wonder Boris had been looking so glum and despondent. However, there is something very strange about their model. I show below a comparison between their modelled cases with those that have actually been recorded so far (October 28). This shows that it is simply their timing that was wrong.
SAGE RWCS modelling of cases (red) compared to recorded cases. The blue curve shows recoded weekly cases
There is an artificial flat phase in their model lasting for about 6 weeks from October to early November. This implies that the modellers had assumed there would have to be a new lockdown period on 1 Oct followed by a relaxation and then a slow surge in cases. Epidemiological models are fairly straightforward unlike these curves . These models are concerned more about social interactions so the scientists are really trying to be more like social engineers testing measures to restrict contact rates.
Note how if we remove the 6 week flat period then the agreement is fairly good. Deaths and cases are ahead of the RWCS because there was no October lockdown. They also appear to be near the peak and the latest ONS infection survey implies that R is slowly reducing. The turnover point would be when the national value of R reaches 1.0. The fall matched to values of R < 1.
Therefore the current tier system is actually working quite well while maintaining economic activity. A full lockdown may well hasten this process but then at the price of causing irrevocable damage both to the economy and to non-covid health care. It may well be that the final death toll from delayed cancer and heart treatment exceeds that due to Covid-19 itself. Boris should therefore wait another week because the current measures may well be enough. New cases are levelling off. Covid deaths today reflect the situation as it was ~3 weeks ago.