The data science approach employed by EEDS is known as time-series analysis. The methodology is well suited to working with high-frequency data that is autoregressive. The case of hourly temperature at Barrow, Alaska, is a good example.
Figure 1. The hourly temperature at Barrow Alaska is highly variable, 1 Jan 1985- 31 Dec 2015
Figure 2. Despite the variability in temperature, the autocorrelations in hourly temperature are highly significant. Controlling for this characteristic makes it possible to extract the CO2 signal from the noise in the data.