Analysis of new research paper tying coronavirus to weather
A new research paper ties temperature and humidity directly to coronavirus transmission rates.
Published Mar 18, 2020 10:07 PM EDT
UPDATE (May 6, 2020): We have another article talking about how it might not be the UV Index as much as the Vitamin-D dose from the sun that could be helping reduce the Coronavirus outbreak in some areas. The image below shows the Vitamin D dose from the sun on the first of each month over the last year, and was obtained from TEMIS.
UPDATE: (April 10, 2020): An article we launched today is a followup to our previous reviews of studies about the weather's effect on Coronavirus and whether or not COVID-19 might be seasonal.
There's a lot in there, but basically the conclusion is that with a novel (new) virus, we just don't have enough data yet. The weather probably has an effect, but other factors are larger so it's difficult to see its effect outside of the noise.
Additionally, a new website from the Copernicus Satellite folks in Europe has launched that shows Coronavirus cases versus temperature and humidity. This is essentially a better version of what I had talked about in my original blog.
UPDATE (March 24, 2020): A new study has been released that says essentially: It's not the heat, it's the humidity, but surmises the effects on the virus may end up being negligible. I also provided research for a new story on AccuWeather.com which says that UV index may also be an important factor, perhaps more important than temperature or humidity.
Original Blog (March 18, 2020): A new research paper has been released which is much more thorough than the data I have reported on previously, and we have a story on AccuWeather.com reviewing it: "New study says 'high temperature and high relative humidity significantly reduce' spread of COVID-19"
Although I am quoted in the story, we couldn't fit all of my analysis of the research paper here. Here's a deep dive:
A research paper entitled "High Temperature and High Humidity Reduce the Transmission of COVID-19" was published on March 9. In the paper, authors Jingyuan Wang, Kai Feng, Weifeng Lv of Beihang University and Ke Tang from Tsinghua University examined coronavirus (COVID-19) cases in 100 cities in China and their correlation to temperature and humidity. They concluded that "The arrival of summer and rainy season in the Northern Hemisphere can effectively reduce the transmission of COVID-19."
Extrapolated to the world, Figure 4 (above) is what they say the transmission rate (R) of COVID-19 will look like this March and this July. As you can see, the Northern Hemisphere will be in much better shape by July, just based on the weather. They even go so far as to say that Tokyo will experience a 48% drop in transmission of the virus by the time the Olympics start in July!
I've reviewed the paper and it seems fairly comprehensive. Its existence on the SSRN website; however, does not mean that it was peer reviewed -- any scientist can submit a paper there and the review process only requires a check for "completeness and relevance." I'm sure its methodology isn't perfect, but I don't see any red flags that make me think that they didn't do their due diligence.
Some of the positive steps that they took to reduce the potential for errors in the China data include:
-
They selected their data before China's intervention to stop the spread of COVID-19 on Jan. 24, because after that the virus growth data is no longer "natural."
-
They accounted for GDP per capita, which they say normalizes differences in health care facilities.
-
They normalized the results for population density.
However, there were some potential issues with the accuracy of their conclusions (which I doubt they could have avoided):
-
For Figure 1, shown above (but not the maps in Figure 4) they used the capitals of the countries as proxies for the weather of the countries, which could be misleading in areas with varied weather.
-
Lacking detailed international data about COVID-19 cases, they used a different timeframe for the rest of the world (Feb. 8 to 29).
-
They assumed the relationship of temperature and humidity to COVID-19 is the same for the rest of the world.
-
Generally in meteorology, we prefer to use dew point, not relative humidity which varies based on time of day, but it's possible in the long-term it wouldn't make a difference.
-
All this still could be a case of correlation without causation and other things occurring at the same time (for example increasing ultraviolet rays).
Lastly, and this is a big one, they assumed that the relative humidity and temperatures from March and July 2019 will be the same as 2020. As discussed previously in this blog, they might have been better advised to use 30-year normals for 2020, as I have suggested in my previous blog.
Interestingly, Figure 3 shows the bulk of the cases in China between the zero- and 10-degree Celcius temperature range, which matches what I found in my previous blog showing January temperatures on a worldwide basis. But, this figure could be more about the likelihood of those temperatures in China in late January, than about a correlation between the Coronavirus and temperature.
Now, meteorologists can pick it up from here and project the weather's effect on the Coronavirus into the summer, although it will also, of course, depend on other factors like each country's place on "the curve" of cases and the success of their response. It could also re-emerge in the fall, as past epidemics have. Here's what AccuWeather and the National Weather Service are predicting for temperatures in the next month and three months:
March is going to be a much warmer month, overall -- over 6 degrees over the I-95 corridor, according to our AccuWeather Enterprise Solutions long-range forecast team. This is great news, and could (if the study above is correct) help reduce the Coronavirus transmission, although you may not see it in the numbers, since testing for the virus is just now starting to roll out in much of the United States.
AccuWeather Long-Range Temperature Forecast for March 2020.
Beyond March, our long-range forecast team believes that the eastern U.S. will continue in an above-normal warm trend through June. All other factors being equal, at least the weather won't be working against us.
AccuWeather Long-Range Temperature Forecast for the April to June 2020 period.
The National Weather Service's forecast for March and April to June also shows higher-than-normal temperatures for most of the country.
NWS forecast temperatures for March 2020.
NOTE: Their forecasts have not been updated since Feb. 20. When they update them this week, I'll replace the ones shown here.
DISCLAIMER: These are my opinions only. See our story for the official review and weather forecast.
Report a Typo
Weather Blogs / WeatherMatrix
Analysis of new research paper tying coronavirus to weather
A new research paper ties temperature and humidity directly to coronavirus transmission rates.
Published Mar 18, 2020 10:07 PM EDT
UPDATE (May 6, 2020): We have another article talking about how it might not be the UV Index as much as the Vitamin-D dose from the sun that could be helping reduce the Coronavirus outbreak in some areas. The image below shows the Vitamin D dose from the sun on the first of each month over the last year, and was obtained from TEMIS.
UPDATE: (April 10, 2020): An article we launched today is a followup to our previous reviews of studies about the weather's effect on Coronavirus and whether or not COVID-19 might be seasonal.
There's a lot in there, but basically the conclusion is that with a novel (new) virus, we just don't have enough data yet. The weather probably has an effect, but other factors are larger so it's difficult to see its effect outside of the noise.
Additionally, a new website from the Copernicus Satellite folks in Europe has launched that shows Coronavirus cases versus temperature and humidity. This is essentially a better version of what I had talked about in my original blog.
UPDATE (March 24, 2020): A new study has been released that says essentially: It's not the heat, it's the humidity, but surmises the effects on the virus may end up being negligible. I also provided research for a new story on AccuWeather.com which says that UV index may also be an important factor, perhaps more important than temperature or humidity.
Original Blog (March 18, 2020): A new research paper has been released which is much more thorough than the data I have reported on previously, and we have a story on AccuWeather.com reviewing it: "New study says 'high temperature and high relative humidity significantly reduce' spread of COVID-19"
Although I am quoted in the story, we couldn't fit all of my analysis of the research paper here. Here's a deep dive:
A research paper entitled "High Temperature and High Humidity Reduce the Transmission of COVID-19" was published on March 9. In the paper, authors Jingyuan Wang, Kai Feng, Weifeng Lv of Beihang University and Ke Tang from Tsinghua University examined coronavirus (COVID-19) cases in 100 cities in China and their correlation to temperature and humidity. They concluded that "The arrival of summer and rainy season in the Northern Hemisphere can effectively reduce the transmission of COVID-19."
Extrapolated to the world, Figure 4 (above) is what they say the transmission rate (R) of COVID-19 will look like this March and this July. As you can see, the Northern Hemisphere will be in much better shape by July, just based on the weather. They even go so far as to say that Tokyo will experience a 48% drop in transmission of the virus by the time the Olympics start in July!
I've reviewed the paper and it seems fairly comprehensive. Its existence on the SSRN website; however, does not mean that it was peer reviewed -- any scientist can submit a paper there and the review process only requires a check for "completeness and relevance." I'm sure its methodology isn't perfect, but I don't see any red flags that make me think that they didn't do their due diligence.
Some of the positive steps that they took to reduce the potential for errors in the China data include:
They selected their data before China's intervention to stop the spread of COVID-19 on Jan. 24, because after that the virus growth data is no longer "natural."
They accounted for GDP per capita, which they say normalizes differences in health care facilities.
They normalized the results for population density.
However, there were some potential issues with the accuracy of their conclusions (which I doubt they could have avoided):
For Figure 1, shown above (but not the maps in Figure 4) they used the capitals of the countries as proxies for the weather of the countries, which could be misleading in areas with varied weather.
Lacking detailed international data about COVID-19 cases, they used a different timeframe for the rest of the world (Feb. 8 to 29).
They assumed the relationship of temperature and humidity to COVID-19 is the same for the rest of the world.
Generally in meteorology, we prefer to use dew point, not relative humidity which varies based on time of day, but it's possible in the long-term it wouldn't make a difference.
All this still could be a case of correlation without causation and other things occurring at the same time (for example increasing ultraviolet rays).
Lastly, and this is a big one, they assumed that the relative humidity and temperatures from March and July 2019 will be the same as 2020. As discussed previously in this blog, they might have been better advised to use 30-year normals for 2020, as I have suggested in my previous blog.
Interestingly, Figure 3 shows the bulk of the cases in China between the zero- and 10-degree Celcius temperature range, which matches what I found in my previous blog showing January temperatures on a worldwide basis. But, this figure could be more about the likelihood of those temperatures in China in late January, than about a correlation between the Coronavirus and temperature.
Now, meteorologists can pick it up from here and project the weather's effect on the Coronavirus into the summer, although it will also, of course, depend on other factors like each country's place on "the curve" of cases and the success of their response. It could also re-emerge in the fall, as past epidemics have. Here's what AccuWeather and the National Weather Service are predicting for temperatures in the next month and three months:
March is going to be a much warmer month, overall -- over 6 degrees over the I-95 corridor, according to our AccuWeather Enterprise Solutions long-range forecast team. This is great news, and could (if the study above is correct) help reduce the Coronavirus transmission, although you may not see it in the numbers, since testing for the virus is just now starting to roll out in much of the United States.
AccuWeather Long-Range Temperature Forecast for March 2020.
Beyond March, our long-range forecast team believes that the eastern U.S. will continue in an above-normal warm trend through June. All other factors being equal, at least the weather won't be working against us.
AccuWeather Long-Range Temperature Forecast for the April to June 2020 period.
The National Weather Service's forecast for March and April to June also shows higher-than-normal temperatures for most of the country.
NWS forecast temperatures for March 2020.
NOTE: Their forecasts have not been updated since Feb. 20. When they update them this week, I'll replace the ones shown here.
DISCLAIMER: These are my opinions only. See our story for the official review and weather forecast.
Report a Typo