Global Warming And Heat Index

I've seen a good bit on temperature increases from global warming and I've put together several visualizations (example), but I haven't seen any clear heat index projections. I went ahead and put some together for the US.

How does global warming affect heat index?

From the definition of heat index, 105 degrees is the cutoff for 'heat stroke is probable with continued activity'. Thus, I used that as my threshold. This is the rough number of days per year where working outside makes a heatstroke probable:

US heat index projections

US heat index

There's a good bit to take in here. A few highlights for the high emissions model:
  • by 2075, Chicago, Omaha, and Cincinnati are projected to be as hot (in terms of heat index)  as present-day Austin, TX heatwaves
  • by 2075, Philadelphia, Pittsburgh, and Detroit are projected to be as hot (in terms of heat index) as present-day Birmingham, AL
  • by 2050, Des Moines and St. Louis are projected to be as hot (in terms of heat index) as present-day Phoenix, AZ
  • by the end of the century, more than 25% of days in much of Texas, Arkansas, and the lower midwest are projected to be probable heatstroke days (compared with <10% today)
Another target I ran is 95 degree heat index. That sits in the 'activity could result in a heat stroke' range. It looks basically the same, but the number of days is much higher:

US heat index

US heat index map

The only new, obvious takeaway to me from that one is that it is projected to be potentially dangerous to work outside for roughly two-thirds of the the year in parts of the south by the end of the century (vs less than half today).

What is heat index?

Heat index is kind of the 'apparent temperature' when resting in a shaded area. For a given hot temperature, it feels hotter at high humidity than at low humidity, primarily because more humidity means less cooling from sweat, and heat index attempts to capture this.

What data did I use?

I used CMIP5 temperature and humidity projections (Berkeley Earth for historic). The temperature projections can be found in a number of places, but the humidity was a bit harder to find and only seem to be available here. I describe how I used the data in a good bit of detail in the daily temperatures section here, but the short summary is:
  • all data were resampled to the same lat/lon grid
  • CMIP5 model results were paired, and the median humidity projections were used for those models where I could not find a humidity projection
  • I grouped the data in 3 year blocks centered at various years (2010, 2025, 2050, 2075, 2095)
  • I used two CMIP5 projections: RCP45 (low emissions) and RCP85 (high emissions). You can see what these mean here
  • relative humidity goes down as temperature goes up (usually) so I used the min humidity
  • I used this equation to calculate heat index:
{\displaystyle {\begin{aligned}\mathrm {HI} &=c_{1}+c_{2}T+c_{3}R+c_{4}TR+c_{5}T^{2}+c_{6}R^{2}+c_{7}T^{2}R+c_{8}TR^{2}+c_{9}T^{2}R^{2}+\\&\quad {}+c_{10}T^{3}+c_{11}R^{3}+c_{12}T^{3}R+c_{13}TR^{3}+c_{14}T^{3}R^{2}+c_{15}T^{2}R^{3}+c_{16}T^{3}R^{3}\end{aligned}}}
T = temperature; R = relative humidity

As I note in the link, my algorithm biases the results towards lower values for heat index since I used minimum relative humidity for the day, so the actual values should be higher than these.


Overall, I like this better as a metric than raw temperatures. If you go off of raw temperatures alone, you get that the southwest is unbearable but the southeast and midwest end up being not too terrible. Factoring in humidity, you see that the problem is much worse and ends up affecting a huge portion of the US population.

No comments:

Post a Comment