Hourly temperature data do not support the views of the Climate Deniers: Evidence from Barrow Alaska

Author: Kevin F. Forbes

Abstract

Survey evidence has indicated that a significant percentage of the population does not fully embrace the scientific consensus regarding climate change. This paper assesses whether the hourly temperature data support this denial. Specifically, this paper examines the relationship between hourly CO2 atmospheric concentration levels and temperature using hourly data from the NOAA-operated Barrow observatory in northern Alaska. At this observatory, the average annual temperature over the 2015-2020 period was about 3.37 oC higher than in 1985-1990. A time-series model to explain hourly temperature is formulated using the following explanatory variables: the hourly level of total downward solar irradiance, the hourly CO2 value lagged by one hour, proxies for the diurnal variation in temperature, proxies for the seasonal temperature variation, and proxies for possible non-anthropomorphic drivers of temperature. A time-series modeling specification is employed to capture the data’s heteroskedastic and autoregressive nature. The model is estimated using hourly data from 1 Jan 1985 through 31 Dec 2015. The results are consistent with the hypothesis that increases in CO2 concentration levels have nontrivial consequences for hourly temperature. The estimated annual contributions of factors exclusive of CO2 and downward total solar irradiance are very small. The model was evaluated using out-of-sample hourly data from 1 Jan 2016 through 31 Aug 2017. The model’s out-of-sample hourly temperature predictions are highly accurate, but this accuracy is significantly degraded if the estimated CO2 effects are ignored. In short, the results are consistent with the scientific consensus on climate change.

Key Points:

1) At NOAA’s Barrow Observatory in Alaska, the annual temperature during 2015-2020 was about 3.37 oC higher than in 1985-1990.

2) Virtually all the upward trend in annual temperature through 2015 can be attributed to higher CO2 concentrations.

3) The model’s out-of-sample predictions are more accurate if the estimated associations between CO2 and temperature are not ignored.