Seminar: Evidence of Causality between Carbon Dioxide Concentrations and Temperature
Date(s): November 19, 2013, 10:30am-12:00pm
Location: Room 4128, William Jefferson Clinton West Building, 1301 Constitution Ave., NW, Washington, DC
Contact: Carl Pasurka, 202-566-2275
Presenter(s): Kevin Forbes (Department of Business and Economics, The Catholic University of America)
Description: Climate change skeptics remain unconvinced that increases in the atmospheric concentrations of carbon dioxide (CO2) have any climate or meteorological implications. In contrast, many climate scientists believe that increases in CO2 concentrations do indeed have meteorological and ultimately climate consequences, but that it is impossible to disentangle these effects from those of other factors for any given individual weather event. This paper contends that it is possible to assess the effect of CO2 and other Greenhouse gases on weather. This paper explores the relationship between CO2 atmospheric concentrations and temperature using hourly CO2 atmospheric concentration data from the Mauna Loa Observatory (MLO) in Hawaii.
The starting point of this paper is the recognition that meteorologists do not explicitly take CO2 induced changes in temperature into account when making weather forecasts. The analysis makes use of day-ahead hourly weather forecast data to control for expected weather conditions exclusive of CO2 considerations. The analysis employs a two-step procedure. In the first step, the issue of functional form is addressed. The results indicate that temperature can be modeled as a nonlinear function of forecasted temperature, forecasted humidity, forecasted dewpoint, forecasted windspeed, and the forecasted probability of precipitation. This evidence of nonlinearity and complexity may help explain why the errors in conventional temperature forecasts are so large. The model is then extended to include CO2 as an explanatory variable taking into account that there is reason to believe that the contemporaneous level of CO2 is an endogenous predictor of temperature.
Using the CO2 augmented results of the first step as a base, an autoregressive moving average process (ARMA) is then modeled. This permits the modeling of the disturbance terms so as to reduce the likelihood of spurious results. The ARMA estimation results are consistent with the hypothesis that changes in hourly CO2 atmospheric concentration levels have implications for ambient temperature. An out-of-sample forecast is then performed using six months of hourly data. Consistent with the existence of a causal relationship, the temperature forecast is shown to be more accurate when CO2 levels are taken into account. The temperature forecast is also more accurate than conventional temperature forecasts for the same location.
Seminar Category: Climate Science