The Nature paper (referenced last entry) determines the threshold for "deadly" days by sifting through and classifying various reported heat wave events in places like France and Southern Australia and comparing them to days with "ordinary weather" in the same place, using a classification technique ("support vector machines") that aims to draw boundaries between types of events based on what appear to be discriminatory boundaries between attributes. In this case it is found with only a fair degree of reliability that heat-attributed deaths generally occur on hot days, with a lesser impact from high humidity. Notably unsurprising conclusions.
By picking points from the chart I developed a simple linear relationship (shown above in green) that emulates the "deadly" boundary. It also appeared to me (as an old psychrometrician) that the boundary appeared to correspond fairly well to a wet bulb temperature of about 26, at least when humidities are 70 percent or more, which is usually the case in Bangkok. Given the broad approximations of the method the study dictates that on days with usual hot-season humidity in the tropics the average daily temperature becomes "deadly" at about 29 degrees.
Readers of this blog might recall that this is the temperature where I turn on the bedroom air conditioner at night. Readers who have some experience with wet bulb and global wet bulb temperatures will be aware that military training and sports practice are thermally acceptable at wet bulb temperatures of 26, heading into precautionary modes only above Twb=26.
