Abstract:
Spectral analysis is appropriate for the analysis of stationary time series and for identifying
periodic signals that are corrupted by noise. These methods are now routinely used in
meteorological studies of temperature. This work considered the application of Singular
Spectrum Analysis (SSA), Maximum Entropy Method (MEM) and Multi-Taper Method (MTM)
to model monthly temperature rates in Ondo State from 1986 to 2015. The work set out to
identify temperature rates, fit a suitable model to the data and make forecasts of future
values. The findings are based on Ondo State alone and on the available data. The data
analyses in this work are the monthly temperature time series in Ondo State from 19
climatological stations, obtained from the Ondo Meteorological Agency (OMA). All the
assumptions associated with statistical tool were tested using the most appropriate technique.
The modeling of temperature in the state was done using SSA, MEM and MTM approach.
Based on the analysis; the study reveals that temperature of the state has increased in recent
time in volume on average temperature of 32.19 degree Celsius with the lowest MT of 10.32
degree Celsius and maximum observation of temperature shows the possibility of average
temperature rising to 42.43 degree Celsius. It was found that using SSA, MEM and MTM give
more information about the underlying series. SSA will be more sensitive to differences in
length, whereas the other two methods will confound length and amplitude of the series.
Among these two, MTM provided more sensitive detection (i.e., detecting lower amplitude
signals), whereas MEM give better frequency specificity.
