Bivariate Discrete Time Series Model for Assessing the effects of Rainfall and Temperature on Road Accidents: The case of Morogoro and Pwani Regions in Tanzania

Abstract

In this article the contributions of weather on the occurrences of road accidents in Tanzania particularly for Morogoro and Pwani Regions is discussed. The correlation between road accidents and weather variables have indirect relationship taking into consideration that the later affects other factors like humans (drivers and other road users), state of vehicles and infrastructure before contributing to the incidence of road accidents. A bivariate discrete time series BINAR(1) model is formulated, and contribution of weather variability conditions in determining and forecasting road accidents is assessed using real data from Tanzania Meteorological Authority and Tanzania Police Force. The results show that, the contribution of weather variables specifically rainfall and temperature on road accidents vary either positively or negatively depending on the particular situation. Therefore, model properties are used to obtain correlation of these factors to road accidents as well as forecasting road accidents. It is not straight forward that increase (decrease) of rainfall or temperature lead to rise (decline) of road accidents. This indicates that, there might be other factors contributing to road accidents that need to be considered for achieving minimal if not zero road accidents. Therefore, we recommend the Government of Tanzania through the Tanzania Police Force to strictly enforce road safety laws and other necessary bylaws as well as providing education about road safety regulations to all road users. This will result to road accidents to be reduced significantly during rain and dry (high temperature) seasons.