Tracking & Managing
Mosquito Breeding Hotbeds
By Philip Harman
​
​In this article, I'll give an overview of how mosquito-borne disease (MBD) outbreaks are predicted using earth observation (EO), and how granular satellite imaging can aid localized mitigation efforts.
Gbongan, Nigeria
​
​How are MDB outbreaks predicted?
Due to the diversity of MBDs and mosquito species, there is an assortment of EO-driven outbreak prediction models. The graphic below is from a review by the University of Erlangen-Nuremberg, and summarizes the methods used by 43 scientific papers on predicting Malaria, Dengue, or West Nile Virus:
​
Feature Prevalence in Epidemelogical Models:
Malaria, Dengue, WNV (source)
​​
As shown, the leading features for predicting MBD outbreaks are temperature, moisture, and elevation. Here's another collection of examples from the NASA Earth Observatory, which shows how temperature and moisture can help predict cases of Rift Valley Fever, Dengue, and Chikungunya virus.
​
NASA Earth Observatory:
Tracking Diseases by Satellite (source)
​
​How can localized mitigation efforts be further enabled?
The examples provided above tend to focus on a national or regional scale, which is useful for warning federal goverments of imminent outbreaks. However, it's also possible to "zoom in" more on the issue (literally). As discussed, the NDVI raster can indicate where mosquito populations will flourish - and consequently, where disease outbreaks will occur.
By making these same observations at the city-level, municipalities could see where mosquito-friendly habitats (ex. marshes) are developing, and take preventative measures.
​
In the example below, I've prepared a Heroku app to show the NDVI view of Gbongan, a city in Western Nigeria (an area that is particularly affected by Malaria). Using a simple thresholding approach, I've shown an approximated view of buildings in red, while the moisture of the remaining areas is shown in blue (darker blue indicates wetter areas).
​
Try zooming on different areas of the map, and look for high-moisture (dark blue) areas.
​
My GitHub; Images from ESA Copernicus Program, via Sentinel Hub EO
\
​
What are some feasible localized mitigation efforts?
First, the issue of insecticides. Researchers have observed a troubling rise in resistance to pyrethroids in recent years. While pyrethroids are still fairly effective, this trend suggests that non-insecticide measures are worth implementing. These include:
Habitat control
A mosquito takes up to 14 days (or less) to go from a newly laid egg to flying, and most species will fly up to 5 km (3 miles) from their hatching location. Meanwhile, the Copernicus satellite program provides new images every 5 days, and can offer as wide of a radius as desired around a point of interest.
​
All that is to say, municipalities could use a tool like the application above to target large breeding grounds within "flight range" of cities and towns. With the relatively low latency of the imagery, these breeding grounds could be systematically eliminated before the larva are able to hatch.
​
Parts of Southeast Asia have shown how effective larval control can be, by draining or filling marshy areas as they appear.
​
Resource distribution
By studying imagery at a city-level, resources like mosquito nets could be distributed to prioritize neighborhoods that are in closest proximity to potential mosquito habitats.
​
Vaccine distributions could be managed in a similar fashion. While many MBDs do not have a human vaccine, livestock vaccines are more widely available, preventing further spread to humans and reducing the economic impact of livestock lost to disease.
​
​
​
Sources:
​