Dynamic Harmonic Regression Modelling with the Minimum or Maximum Temperatures of Umuahia

Abstract

Air surface temperature is one of the most important environmental factors that serve as a climate change indicator. A separate analysis on the minimum and maximum temperatures would be more informative than on the average temperature. The data used spanned over 30 years (1988 - 2017) separately available for minimum and maximum temperatures. Dynamic harmonic regression methods were successfully applied to the minimum/maximum temperature series where the error components were allowed to correlate but with innovations that follows a WN distribution. Minimum temperature series was detected to have a linear trend, double seasonal periods of 6 months (semi-annual) cycle and 12 months (annual) cycle, this multiple seasonality is usually ignored by most traditional time series methods, however, this was adequately handled using Fourier series. The increase in mean surface air temperature in Umuahia metropolis and its environs is believed to be due to the increase in the minimum temperatures rather than in the maximum temperatures, the minimum temperature was found to be increasing at the rate of 0.024oC yearly and 0.7oC over 29 years period, whereas the maximum temperature was significantly stable over the said period.