Will read the whole study soon, and add thoughts below, but recent claims of a spike vehicles parked at hosptials in Wuhan in October seem VERY flimsy. Especially since they included this image where the off-nadir angle reveals a whole additional parking area + new carparks 🙄 
This problem seems to persist in pretty much every example they use. The data is noisy as anything, and actually says more in the drop of numbers after the first confirmed diagnosis (2nd dotted line) than the slight rise in the months preceding.
Moreover, the 1st dotted line represents an increase in internet searches about diarrhoea, a symptom in less than 1/3 of symptomatic patients, without a simultaneous rise in the number of searches about coughing, a symptom of >80% of cases, which doesn't rise until mid-late Dec.
The manuscript explains this by saying "this finding lines up with the recent recognition that gastrointestinal (GI) symptoms are a unique feature of COVID19 disease and may be the chief complaint of a significant proportion of presenting patients"
citing a journal that says 34%
citing a journal that says 34%
Amazingly the authors suggest that their search term analysis may be revealing new "missed early signals" of COVID, as though google searches could tell you something about the disease that 7.2mil confirmed cases hasn't.
Which is very putting the cart before the horse.
Which is very putting the cart before the horse.
The authors also suggest that looking for influenza-like symptoms may be missing milder cases, whereas the study on COVID and GI symptoms they cited before explicityly says that GI symptoms are more prevalent, and that <3% of patients showed GI symptoms without respiratory ones.
They talk about the limitations of satellite imagery caused by shadows, but not the bigger issue of buildings themselves blocking the view.
"Exactly overhead" hints at this, but is false.
In 2018 and 2019 I could find exactly 0 satellite images of Wuhan taken from +-5° overhead.
"Exactly overhead" hints at this, but is false.
In 2018 and 2019 I could find exactly 0 satellite images of Wuhan taken from +-5° overhead.
All in all this study strikes me as an investigation into a very certain hypothesis, which led the analysis and ended up being published desite extremely inconclusive data being found. This is not the way you want to look into these issues, and I'd even say is counterproductive.
H/T @HarelDan who also annotated the first image used here
He also mentions that the one example shown that doesn't suffer from buildings blocking the view has a difference of about 100 cars.
He also mentions that the one example shown that doesn't suffer from buildings blocking the view has a difference of about 100 cars.
... I'm thinking of doing a similar quick study that tries to imply things from vehicle counts (not about Wuhan), if I end up pulling it together, you'll see how VASTLY different my approach is to that of this study.
