Abstract: Using mosquito and arbovirus data from the Connecticut Agricultural Experiment Station’s (CAES) statewide mosquito and arbovirus surveillance program, we developed quantitative methods that project arbovirus risk at the scale of surveillance sites within a network as well as in un-sampled spaces that surround sites within a network. For among site projections, we utilized population synchrony methods which defined spatiotemporal relationships of mosquito and arbovirus collections among sites as well as investigated the climatic and environmental variables associated with high synchrony estimates. For risk projections into un-sampled space, we used machine learning methodologies to first develop predictive algorithms of Culex pipiens or Culiseta melanura collections based on climate, land cover, and prior monthly collections; these algorithms were then nested within a predictive model of West Nile virus or eastern equine encephalitis virus detection probabilities. The over-arching goal of this research is to develop interactive, online risk maps which can be released to the public by CAES during a surveillance season. The predicted utility of such risk maps is that they will allow users to estimate arbovirus risk at sites within the surveillance network which may test negative for a virus or in locations not explicitly sampled by the surveillance network.
