West Nile Virus (WNV) is a mosquito-borne disease. It was first discovered in the West Nile District of Uganda in 1937. According to the reports from the Centers for Disease Control and Prevention, WNV has been found in Africa, the Middle East, Europe, Oceania, west and central Asia, and North America. Its first emergence in North America began in the New York City metropolitan area in 1999. It is a seasonal epidemic in North America that normally erupts in the summer and continues into the fall, presenting a threat to environmental health. Its natural cycle is bird-mosquito-bird and mammal. Mosquitoes, in particular the species Culex pipiens, become infected when they feed on infected birds. Infected mosquitoes then spread WNV to other birds and mammals including humans when they bite. In humans and horses, fatal Encephalitis is the most serious manifestation of WNV infection. WNV can also cause mortality in some infected birds.
The spread of WNV has shown unique distribution patterns in different regions [1–5]. Environmental determinants, such as the presence of suitable habitats, temperatures, and climates, play important roles in WNV dissemination in North America [6, 7]. Mosquito Culex species appear to prefer some land use and land cover (LULC) types (e.g., wetlands and specific grasslands) than some others (e.g., exposed dry soils). Mosquitoes in the canopy site are believed to possess more infections than those in subterranean areas and on the ground . Wetlands and stormwater ponds, especially those under heavy shade, provide an ideal environment for mosquito settlement. Ponds with plenty of sunshine and a shortage of vegetation are believed to be a poor environment for mosquito development . WNV dissemination is found to be significantly related to average summer temperatures from 2002 to 2004 in the USA. .
Field and laboratory records of entomological and ecological observations have been used to examine how natural environmental constraints, such as water sources and climatic parameters, contribute to the transmission of WNV [11, 12, 9]. Doham and Turell (2001) found that the infection rates of WNV in mosquitoes are lower at cooler temperatures than when these vectors were maintained at warmer temperatures. The infection rates start to increase after one day of incubation at 26°C. WNV dissemination begins more rapidly in mosquitoes settled at higher temperatures than in mosquitoes maintained at cooler temperatures . Gingrich et al. (2006) detected a bimodal seasonal distribution of mosquitoes with peaks in early and late summer in Delaware in 2004, and that mosquitoes are attracted to ponds with heavy shade and low slopes.
Remote sensing (RS) and geographic information system (GIS) technologies have been extensively applied in public health studies and related issues such as urban environmental analysis [13–15, 6, 5, 16–20]. These technologies have been applied to research diverse epidemiological issues, such as parasitic diseases and schistosomiasis using RS and GIS as exclusive sources of information for studying epidemics. The accessibility of multi-temporal satellite imagery effectively supports the study of epidemiology . Ruiz et al. (2004) found that some environmental and social factors contributed to WNV dissemination in Chicago in 2002 by using GIS technologies and multi-step Discriminant Analysis. Those factors included distance to a WNV positive dead bird specimen, the age of housing, the intensity of mosquito abatement, the presence of vegetation, geological factors, and demographic factors such as population age, income, and race. Multiple mapping techniques were compared for WNV dissemination in the continental USA . The results indicated that each mapping technique emphasized certain WNV risk factor(s) due to the differences in modeling assumptions, statistic treatment, and error determination. There was no single model performing better than all others. Cooke III et al. (2006) estimated WNV risk in the state of Mississippi based on human and bird cases recorded in 2002 and 2003 with the creation of avian GIS models. The results indicate that high road density, low stream density, and gentle slopes contributed to the dissemination of WNV in Mississippi. GIS and spatial-time statistics were applied for a risk analysis of the 2002 equine WNV epidemic in northeastern Texas . A total of nine non-random spatial-temporal equine case aggregations and five high-risk areas were detected in the study area. Ruiz et al. (2007) further examined the association of WNV infection and landscapes in Chicago and Detroit using GIS and statistical analysis. Their results show that higher WNV case rates occurred in the inner suburbs where housing ages were around 48–68 years old with moderate vegetation cover and population density.
Many valuable studies have documented the effects of environmental and socioeconomic factors on the spread of WNV. Despite this, a long-term study of these effects using data with high temporal resolution has yet to be undertaken. This study develops a multi-temporal analysis of the relationship between environmental variables and WNV dissemination using an integration of remote sensing, geographic information systems (GIS), and statistical techniques. The specific research objectives are to identify the spatial patterns of WNV outbreaks at different years and seasons in the city of Indianapolis, USA, to examine the relationships of WNV dissemination and environmental variables, and to investigate the temporal variations of the relationship. Through spatio-temporal analyses, it is possible to identify and explain the temporal outbreaks of WNV and the high-risk areas in the study area.
Indianapolis is a typical Midwest city lying in the flat plain and has a temperate climate without pronounced wet or dry seasons. Therefore, this study can offer not only valuable information for public health prevention and mosquito control, but also provide a testimony for the WNV spread in the other regions of the Midwest USA and beyond.