Improved comfort scores

Back in July 2016 I first laid out the logic and methodology for computing "comfort scores" based on the deviation of temperatures on each day from the range considered ideal for human comfort, and I expanded upon this more practically in June 2017. (Charts for the most-recent periods are found in the Recent Weather section.) The true discomfort of low temperatures is estimated from the wind chill, and of high temperatures from the heat index. I've since made several refinements to the calculation. These include introducing a diurnal cycle to the range of ideal temperatures, so that now 64-70 F is considered ideal at night, and 70-80 F is considered ideal during the day. I also have run the calculation for North American Regional Reanalysis data to get higher-resolution results for the United States. The most-substantial change, however, is the inclusion of precipitation rate as an additional variable, along with the original ones of temperature, humidity, and wind speed. This means the discomfort is calculated as a sum of contributions from heat, cold, and wetness. From experience we all know that the effect of precipitation on comfort varies with temperature — cold rain is much more unpleasant than warm rain, and at very hot temperatures rain is welcome, as long as it's not very heavy. These relationships are captured in the below diagram illustrating the bivariate nature of the precipitation component of the discomfort score.

Illustration of the discomfort score as a bivariate function of both precipitation rate and temperature. Units are English for ease of applicability in the United States.

With this new definition, I recalculated the scores for the entire world using NCEP reanalysis, and for the continental US using NARR reanalysis. In the galleries below are the new climatologies for discomfort scores (red being most uncomfortable, blue most comfortable), encompassing the annual average as well as seasonal averages for winter, spring, summer, and fall, for the world (top) and the US (bottom). Key features include the complete unsuitability of Antarctica for human comfort, the appeal (climatologically speaking) of the subtropical ocean gyres, and the relative comfort of much of the tropics, especially those areas at several thousand feet elevation such as southern Africa, Central America, and Southeast Asia. Nonetheless, the hottest and most-humid areas (the northern Indian Ocean, western Pacific, and coastal Arabia) see summertime discomfort levels much greater than those elsewhere in the tropics, and in fact they approach those of Antarctica in austral summer (but are nothing like those in austral winter). Due to the inclusion of precipitation as discomfort-score factor, there is a greater value placed on dryness, resulting in the improved performance of the western US relative to the eastern. This makes the Southwest and Central Valley the most comfortable places on an annual-average basis, and even in summer they are much more comfortable than the humid Gulf of Mexico and Gulf of California.

I also used the revised discomfort-score climatology to compute time series over the period of available reanalysis data, 1979-2016. The next figure shows the trend in contributions to globally averaged discomfort for heat and cold. As the world has gotten warmer, discomfort has come more and more from heat rather than cold. Individual hot years also stand out, such as the El Ninos of 1983, 1998, and 2016. Note that discomfort is a nonlinear function of temperature due to the compounding effects of humidity and wind speed, as well as the threshold effects introduced by the calculation methodology. Regardless, a clear change is evident since about 1995, and these trends will likely continue as more locations experience uncomfortable heat while simultaneously experiencing less uncomfortable cold.

The last figure shows the same information, but broken down into latitudinal averages. Shown are percentiles relative to the 1979-2016 average for that latitudinal band; for example, a year falling in the 90th percentile means it was more uncomfortable overall than 90% of years in that location. Discomfort in the tropics and subtropics is primarily a function of heat, whereas in the mid-latitudes and polar regions it is a function of cold. Therefore, a trend toward greater comfort is expected poleward of 30 deg and toward less comfort equatorward of that — and that's exactly what is shown here. The last two years have exhibited record discomfort in the tropics and record comfort at the poles, which is again a trend very likely to continue.