There's been an ongoing thread about the reliability of ground temperatures in EPW files, with the common lament that these temperatures are based on an old study by Kusuda and Achenbach from more than 50 years ago (1965 as an ASHRAE paper, 1973 as a NBS report). The Kusuda/Achenbach model is a one-harmonic model, i.e., sine curve, with the coefficients for amplitude and lag based on data for 28 locations in the US. As used in all the simulation programs and weather files, this is reduced to a sine curve with amplitude and lag based solely on the annual air temperature profile (also reduced to a sine curve) and the depth below the surface, with the soil diffusivity held constant.
I was just at the ASHRAE Conference in Chicago (Jan 21-24, 2018) and happened to see a technical paper that describes a much improved two-harmonic model that also varies the soil conditions depending on the Koppen-Geiger climate classification. I thought such a breakthrough warrants starting a new thread instead of merely as a comment to the ongoing thread.
