Simulating Mosquito Habitats as an Element of Climate-informed Disease Forecasting

Mosquito-borne disease incidence is dependent on climate. Disease risk assessment studies often incorporate climate via meteorological variables (e.g., rainfall, temperature, humidity), but what is less studied are subseasonal climatic impacts on vector breeding habitats—the water-filled containers and puddles that house the vector’s aquatic early life stages, and in which vector growth can either be hindered or promoted based on variability in water level and temperature. Our work investigates the subseasonal climate sensitivity of these mosquito habitats as a key intermediate step in the climate-disease process chain; how does variability in climate translate to variability in habitat suitability?

We approach this question by simulating the breeding habitats of Aedes aegypti and Aedes albopictus, the mosquito vectors of pathogenic viruses causing chikungunya, dengue, yellow fever, and Zika. These urban-adapted Aedes species often breed in artificial containers (e.g., water tanks, flower pots, discarded tires), which we model using the energy balance container model WHATCH’EM (NCAR). WHATCH’EM simulates the hourly temporal evolution of water height and temperature within a container habitat based on user-specified parameters (such as container dimensions, shading, thermal conductivity) and climate inputs (timeseries of rainfall, air temperature, relative humidity).

We characterize the climate sensitivity of these simulated container habitats in terms of: (a) sensitivity to interannual climate variability, assessed by WHATCH’EM runs with several different years of climate data, and (b) sensitivity to climate forecast error, assessed by WHATCH’EM runs with retrospective subseasonal-to-seasonal meteorological forecasts and corresponding historical climate data. Via the former we assess whether the climate sensitivities of (simulated) vector habitats are more nuanced than the well-known seasonal cycles of disease incidence, while the latter provides insight into the value of forecast data for such simulations (a tradeoff between additional lead time and additional uncertainty). From this characterization of climate sensitivities we assess the added value of this process-based modeling approach, informing potential integration of such habitat modeling into mosquito-borne disease forecasting systems.