Maximum Likelihood is generally considered as one of the most popular methods to estimate the parameters of spatio-temporal covariance functions. There are two major challenges, in this concern, uncertainty to estimate a great number of parameters and computational time when the dimensions of time or space is increased. In this paper we use Composite and Weighed Composite Likelihood functions for reducing the computational time and try to combine these methods with Bees Algorithm (BA) for finding optimal values of spatio-temporal covariance function. Next, the accuracy and computational time of these methods are compared based on simulation study. Finally, the new combined method is used to explore the spatio-temporal correlation structure of monthly temperature data provided by the Illinois Climate Network.
