Mosquito trap counts are heavily influenced by environmental factors such as temperature and precipitation. However, some important geographic factors, such as land use and elevation of a particular site, are often either not recorded or simplify not observable. This is a major issue in building a predictive model for the mosquito trap counts over time across a particular region. The collective impact of all unobservable factors for one particular site is estimated by a hidden dimension method. Application to mosquito trap counts in Peel Region has shown that our model can significantly improve the modeling accuracy of the generalized linear model. This method may provide a significantly better characterization of the spatiotemporal distribution of mosquito (Diptera: Culicidae) abundance in areas with green lands or open spaces.