High resolution West Nile virus risk mapping using satellite-derived land surface temperature data

Anna Boser and Ashley E. Larsen, Bren School of Environmental Science and Management, University of California Santa Barbara, Santa Barbara, CA, Daniel Sousa, Jet Propulsion Laboratory, California Institute of Technology, Andrew J. MacDonald, Earth Research Institute, Bren School of Environmental Science and Management, UC Santa Barbara, Santa Barbara, CA

Background/Question/Methods
Mosquito-borne diseases threaten over 80% of the world’s population, and are increasing in intensity and geographical range with environmental change. Understanding the risk profile of mosquito-borne illness is instrumental for mitigation efforts, but limitations on the spatial and temporal resolution of available data leads to a gap in fine-grained understanding of where and when risk is present. The risk profile of West Nile virus (WNV), the most prevalent mosquito-borne illness in California, is composed of well-understood relationships between many aspects of WNV ecology and temperature, its main abiotic determinant. Here, we leverage the established relationship relating air temperature to mosquito biting rates and West Nile virus transmission probabilities and fine resolution land surface temperature (LST) measurements from the Ecosystem Spaceborne Thermal Radiometer Experiment (ECOSTRESS) (2018-2020) to create fine resolution maps of WNV risk over different times of day. To do so, we first use the LST measurements to model air temperature at a 70m resolution and then apply the laboratory derived equations to resolve biting and transmission rates. We then evaluate the ability of this approach to tease apart previously uncharacterized spatio-temporal variations over different land cover types by utilizing high quality agricultural data available in the greater Bakersfield, CA area.
Results/Conclusions
We find that high resolution air temperature maps, and thus WNV risk maps, can be successfully modeled from ECOSTRESS LST measurements (R2 = .85). The fine spatial resolution combined with the different times at which such maps are available allow us to resolve diurnal variations in mosquito biting rates and WNV transmission rates over different land cover types. We find that agricultural areas are cooler as compared to urban areas throughout the diurnal cycle. Due to the nonlinear nature of the epidemiology of WNV in relation to temperature, this translates to lower biting and transmission rates in agricultural areas at night and higher ones during the day. Additionally, a clear stratification of temperatures across different crop types is observed during the day, leading to differences in WNV transmission probabilities. Uncultivated fields display the highest temperatures, and lowest transmission probabilities. These insights show that high resolution LST data can give us insights into WNV ecology and risk at new resolutions, which presents itself as an opportunity to devise more targeted interventions and predictions under environmental change.