MODELLING THE DYNAMIC RELATIONSHIP BETWEEN RAINFALL AND TEMPERATURE IN KASSENA- NANKANA MUNICIPALITY

Abstract

Rainfall and temperature are important climatic inputs for agricultural production especially in the context of climate change. The study employed a Vector Autoregression (VAR) model to examine the dynamic relationship between rainfall and temperature time series data in Kassena-Nankana Municipality, collected from the Navrongo Meteorological Service which spanned from January 2000 to December 2012. The findings revealed that; the coefficient of variation is low in temperature data and high in rainfall data. The linear and exponential trends model was the best to fit rainfall and temperature respectively. VAR model favoured VAR at lag 5 which indicated bi-directional causation from rainfall to temperature and from temperature to rainfall. A univariate ARCH-LM test and Ljung-Box test revealed that the model is free from conditional hetroscedasticity and serial correlation respectively. The Impulse Response Function and the Forecast Error Decomposition were further used to interpret the VAR model. The magnitude of the forecast uncertainty of rainfall that is accounted for by temperature innovations was 13.05% and that of temperature accounted for by rainfall innovations was 9.05%. The study concluded that there is a bi-directional relationship between rainfall and temperature, indicating that rainfall is useful in explaining an appreciable amount of the forecast uncertainty in temperature, and vice versa. Thus modelling rainfall and temperature together in the study area will further improve the forecast of rainfall and temperature respectively. The research strongly recommends that the results should be considered so as to help in planning purposes and appropriate applications in agricultural activities.