13.2.1 — Disease Intelligence — maturity: live
Vector-Borne Disease Risk Mapping
Continuously mapping mosquito, tick and sandfly breeding habitat from orbit to predict where malaria, dengue, Zika and Leishmaniasis will strike next.
Satellite-derived land surface temperature, vegetation indices, and soil moisture data let national health ministries pinpoint where mosquitoes, ticks, and sandflies will thrive before an outbreak ignites.
Ministries of health are perpetually reactive: outbreaks are confirmed weeks after transmission has already peaked, when vector populations have already exploded and the window for targeted intervention has closed. The conditions that drive those populations — standing water, humid vegetation, urban heat islands, deforested edge habitat — are all observable from orbit at resolutions that ground-based surveillance cannot match at national scale. A sovereign satellite stack turns that environmental signal into a forward-looking risk map updated every few days, not every few months.
The satellite stack combines multispectral imagery for vegetation and water indices (NDVI, NDWI, surface temperature), synthetic aperture radar for soil moisture and sub-canopy standing water invisible to optical sensors, and meteorological data ingested from partner weather satellites. Fused at pixel level, these inputs feed species-specific habitat suitability models — Anopheles gambiae behaves differently from Aedes aegypti — producing gridded risk scores down to 30-metre resolution. That granularity lets vector-control teams prioritise larviciding and indoor residual spraying at ward or village level rather than blanketing entire provinces.
The operational outcome is a shift from crisis response to anticipatory public health. A national disease intelligence authority running this system can pre-position insecticide stocks and community health workers two to four weeks before transmission season peaks, based on objectively derived risk scores rather than anecdote. Countries that depend on commercially licensed risk products from foreign vendors cannot adjust model parameters to their own endemic species mix, cannot guarantee data continuity during diplomatic friction, and cannot integrate classified population-movement data that sharpens the exposure estimate. Sovereign control closes all three gaps.
Frequently asked
Which satellite data inputs are most useful for predicting malaria risk?
The core trio is land surface temperature (LST) from MODIS or VIIRS, normalised difference vegetation index (NDVI) from Landsat or Sentinel-2, and surface water extent from SAR (Sentinel-1 or ICEYE). These proxies capture mosquito development temperature thresholds, breeding habitat greenness, and standing-water persistence — the three dominant environmental drivers of Anopheles suitability. Rainfall estimates from GPM IMERG improve the model further by predicting ephemeral pool formation.
Why shouldn't a government simply buy risk maps from a commercial provider like Planet or Spire?
A commercially procured map is a snapshot licensed under terms the vendor can change, restrict, or price up at contract renewal. When a government owns the downstream processing pipeline — even if feeding on open Sentinel or Landsat data — it retains the analytical capability, the historical archive, and the freedom to integrate classified population or military-movement data that no commercial vendor should hold. During a declared public health emergency, speed and control of the data product matter more than cost savings on the imagery itself.
How often do risk maps need to be refreshed to be operationally useful?
Entomological evidence suggests that Anopheles habitat changes significantly on a 5–10 day cycle following rainfall events, so weekly updates are the minimum meaningful cadence for intervention targeting. Daily revisit constellations (Planet SuperDove, Satellogic) make this achievable, while the free Sentinel-2 constellation offers 5-day revisit at 10 m resolution as a cost-free baseline. Risk indices for slower-moving vectors like sandflies (Leishmania) can tolerate monthly updates.
Can a low-income country realistically build this capability indigenously?
Yes, at the analytics layer — which is the highest-value part. Open data from USGS Landsat, ESA Sentinel, and NASA MODIS is free; processing using Google Earth Engine or open-source tools (SNAP, QGIS, R) costs nothing in licensing. A nation needs perhaps 4–6 trained earth-observation analysts embedded in the national public health institute, supported by a sovereign cloud or government data centre. Owning the satellite constellation itself is a second-order ambition; controlling the analysis pipeline is the immediate sovereign priority.
How does this differ from the outbreak hotspot detection application?
Vector-borne disease risk mapping is prospective and environmental — it identifies where conditions favour vector survival and reproduction before human cases occur. Outbreak hotspot detection is retrospective and case-based — it clusters confirmed or suspected cases to find active transmission foci. The two are complementary: risk maps guide pre-emptive vector control, hotspot detection triggers rapid case response. Ideally, both feed the same national health dashboard.
What accuracy should a government expect from a satellite-based risk model?
Peer-reviewed studies in sub-Saharan Africa report area-under-curve (AUC) values of 0.75–0.88 for satellite-driven malaria risk models, meaning they correctly rank high-risk areas significantly better than chance but are not perfect. Accuracy degrades during cloud-contaminated seasons, in data-sparse regions, and when applied outside the geographic area used for model training. Governments should treat outputs as decision-support tools to prioritise field investigation, not as definitive maps that replace entomological surveillance.
What role does the WMO and EUMETSAT data infrastructure play?
WMO's Global Telecommunication System and EUMETSAT's data services provide near-real-time meteorological inputs — particularly precipitation, humidity, and surface temperature — that are essential driver variables for vector-habitat models. EUMETSAT's Meteosat Third Generation (MTG) satellites, launched from 2023, deliver 15-minute full-disk thermal imagery over Africa and Europe that substantially improves the temporal resolution of LST inputs to disease models.
Are there international agreements that obligate data sharing for disease surveillance?
The WHO International Health Regulations (IHR 2005) oblige member states to notify WHO of potential public health emergencies of international concern (PHEIC) and to share relevant epidemiological data, which increasingly includes environmental and geospatial datasets. However, the IHR contains no specific mandate to share satellite imagery, leaving data-sharing largely voluntary and subject to bilateral negotiation. This gap is one reason sovereign analytical capability — rather than reliance on a neighbour's goodwill — is essential.