The development of a spatially resolved retrospective ensemble forecast model of West Nile virus in Long Island, New York

Abstract

The development of a spatially resolved retrospective ensemble forecast model of West Nile virus in Long Island, New York Nicholas DeFelice, Meytar Sorek-Hamer, Scott R Campbell and Krishna Vemuri West Nile virus (WNV) emerged in the western hemisphere in 1999, and has established itself as the leading cause of domestically acquired arboviral disease in the United States. Though the transmission of WNV exhibits a pronounced sensitivity to a complex seasonal ecology along with hydrological and meteorological conditions, our ability to predict the timing, duration, and magnitude of local WNV outbreaks remains limited. In Long Island, New York, annual peak estimates of WNV infected mosquitos have ranged from 1 in 300 to 1 in 50 resulting in anywhere from 0 to 24 reported human WNV cases, respectively. Here we report the development of a spatially refined model that uses ECOsystem Spaceborne Thermal Radiometer Experiment on Space Station (ECOSTRESS) data to capture the variability in physical environmental factors (e.g., temperature and hydrology) along with a compartmental model describing WNV transmission dynamics, to retrospectively forecast WNV outbreaks in Long Island, New York. The inclusion of ECOSTRESS’ high spatial (70m) and temporal (revisits every 4 days) resolution data will allow us to capture changes in the micro-ecosystem. Ensemble simulations with the WNV model are iteratively optimized using data assimilation methods and observations of human incidence, mosquito abundance, and mosquito infection rates. The ECOSTRESS meteorological and hydrological indicators are included in the core model structure to better constrain WNV amplification and transmission dynamics. The model-inference system can be used to better understand the spatial variability of the outbreak and estimate the relationship between zoonotic amplification and human outbreaks. This work represents an initial step in the development of a statistically rigorous system for a spatially resolved real-time forecast of seasonal outbreaks of West Nile virus.