Environmental characterisation of an urban area to assess the distribution of Aedes aegypti monitoring system

Aedes aegypti is the main vector of dengue, chikungunya and zika viruses, which together have resulted in the highest rate of disease and mortality among emerging and/or re-emerging viruses in the Americas. Vector surveillance is a key tool for prevention and control of these diseases. In this context, a proper distribution of sensors within a city will provide timely and precise information to guide public health actions. Under the assumption that environmental variability will determine different probabilities of mosquito presence and activity, our objective was to characterise the urban coverage of Córdoba city at neighbourhood and census tracts levels in order to determine an optimal distribution for the ovitrap network based on the environmental variability. To this aim we first classified very high resolution (VHR) satellite imagery following an object based (GEOBIA) approach. Then, we estimated several landscape metrics for neighbourhood and census tracts polygons and performed a k-mean clustering to determine groups of environmentally similar polygons over the city. After different tests, we defined four environmental clusters for the census tracts and three for the neighbourhoods. Finally, we distributed 150 ovitraps over the city based on the environmental groups defined and compared this distribution with the one used by the Health Ministry, a random one and, a systematic one. It was observed that the arbitrary distribution is the least environmentally representative of the city both for neighbourhoods and census tracts. Instead, the ovitrap distribution stratified by clusters at census tracts level was the best option as it properly covers the environmental variability detected over the city.