Tick-borne encephalitis habitat range expansion from vegetation and climate data
Ixodes tick populations are advancing into higher latitudes and elevations as winters shorten. Combining MODIS land surface temperature seasonality, VIIRS snow-cover duration, and Sentinel-2 forest-edge density lets health agencies map where TBE risk is arriving before the first human cases confirm it.
Sensors
- MODIS MOD11A2 (Terra/Aqua LST): 1 km spatial resolution, 8-day composites of land surface temperature. Used to characterise the length and thermal intensity of the active season for questing ticks, particularly the accumulation of degree-days above the ~7°C threshold at which Ixodes ricinus becomes active.
- VIIRS VNP10A1 (Snow Cover): 500 m daily snow-cover fraction from the Suomi-NPP and NOAA-20 platforms. Snow-free season duration is a direct proxy for the period during which ground-level tick activity is possible; multi-year trends in this product reveal where that window is lengthening.
- Sentinel-2 MSI: 10 m multispectral imagery at roughly 5-day revisit (two satellites combined). Used to map forest-edge density and compute NDVI, both strong predictors of Ixodes habitat suitability. Forest-edge length per unit area is calculable at meaningful ecological resolution; cloud cover remains a constraint at high latitudes in autumn and winter.
- ERA5 Reanalysis (ECMWF): ~31 km gridded climate reanalysis back to 1940, providing minimum winter temperature trends. Confirms whether observed LST warming signals represent a persistent shift rather than interannual noise, and supports degree-day accumulation modelling at landscape scale.
Why the forest edge is the front line
Ixodes ricinus, the primary European vector of tick-borne encephalitis virus, is not a deep-forest species. It concentrates at the boundary between woodland and open ground, where humidity stays high enough for survival but host animals, deer, rodents, and humans, pass through regularly. Sentinel-2 at 10 m resolution can resolve these ecotones in a way that older moderate-resolution sensors cannot. A 30 m pixel blurs the edge; a 10 m pixel places it.
Forest-edge density, calculated as metres of woodland boundary per hectare of landscape, is among the strongest landscape-level predictors of tick encounter rates in published habitat suitability studies. Combining this with NDVI-derived understorey density gives a structural habitat score that can be updated seasonally as leaf-out and senescence alter the optical signal.
What snow duration tells you that temperature alone does not
A warming trend in mean annual temperature is too coarse a signal to schedule a vaccination campaign. What matters operationally is when the ground becomes snow-free at a specific location, because that is when questing ticks become active. VIIRS VNP10A1 provides daily 500 m snow-cover fraction, and from a multi-year archive you can calculate the date of consistent snow clearance for each pixel, along with how much earlier that date is arriving compared with a baseline decade.
In parts of Scandinavia and the Baltic states, the snow-free season has extended by two to four weeks over recent decades, according to published analyses of satellite snow-cover records. That is not a marginal shift. It means tick activity windows that previously ended before late-autumn school outdoor activities now extend into them. The same logic applies at the upslope margin in the Alps and Carpathians, where elevation once provided a reliable thermal barrier.
One honest caveat: VIIRS snow cover struggles in dense forest canopy, where the canopy itself masks ground snow. In heavily forested grid cells, snow-free date estimates carry larger uncertainty and should be cross-checked against ERA5 surface temperature records.
Thermal accumulation and the degree-day model
Tick development and questing activity are temperature-gated. The commonly used lower threshold for Ixodes ricinus activity is approximately 7°C land surface temperature; larval and nymphal development rates accelerate non-linearly above this point. Accumulating degree-days above this threshold from MODIS MOD11A2 8-day LST composites produces a seasonal phenology index that can be mapped at 1 km and compared year-on-year.
The practical output is a projected activity-window map: for each pixel, the expected first and last dates of significant nymphal questing activity, which is the life stage most responsible for human TBE transmission. Nymphs are also the hardest to detect on the body, making the timing of public-health warnings particularly important. ERA5 minimum winter temperature trends provide the longer-term context, distinguishing sites where warming is crossing a habitat-viability threshold for the first time from sites where ticks have always been present but are now active longer.
A limit worth stating plainly: LST is skin temperature, not air temperature at 10 cm above the litter layer where ticks actually quest. The relationship between the two is consistent enough to be useful but introduces uncertainty, particularly on clear-sky days with high solar loading on south-facing slopes.
Translating habitat maps into vaccination campaign geometry
A habitat suitability model produces a continuous risk surface. A health ministry needs discrete decisions: which districts to prioritise for TBE vaccination roll-out, and when to run public-awareness campaigns before the spring nymphal peak. The translation requires overlaying the risk surface with population data and, critically, with records of where vaccination coverage is currently low.
The geographic expansion front is the actionable part. Districts that crossed from low to moderate habitat suitability within the past five to ten years are the ones where vaccine-naive populations are most exposed. These are identifiable from the satellite archive without waiting for surveillance case reports, which typically lag exposure by weeks and are subject to healthcare-access bias. Mountainous districts where the 7°C isotherm has recently climbed above 600 m are a clear example of this category across Central Europe.
Satellize can structure this output as an annually updated GIS layer showing habitat suitability change by administrative unit, ranked by the combination of suitability gain and unvaccinated population, similar in logic to the prioritisation approach used in the Tonga crop-estimation programme, where spatial analytics drove resource allocation decisions across dispersed geography.
Honest limits of the approach
Satellite data cannot detect ticks. It detects the environmental conditions that published field studies associate with tick presence and activity. The model is only as good as the entomological training data used to calibrate it, and those datasets are geographically uneven. Northern Scandinavia and the Caucasus have thinner field-survey coverage than Central Europe, which introduces spatial bias in model confidence.
Cloud cover is a persistent problem for Sentinel-2 at high latitudes, particularly in the autumn window when understanding the end of the activity season matters. Multi-year compositing mitigates this but reduces temporal precision. MODIS LST has its own cloud-masking gaps; 8-day composites help but do not eliminate them. Any operational system should carry explicit uncertainty bands on predicted activity-window dates rather than presenting them as precise forecasts.
Finally, tick-borne encephalitis transmission also depends on host animal density, particularly roe deer and bank voles, which satellite data cannot directly measure. A habitat suitability map is a necessary input to risk assessment, not a sufficient one.
Typical figures
| Habitat structural resolution (Sentinel-2) | 10 m per pixel |
| LST resolution (MODIS MOD11A2) | 1 km, 8-day composite |
| Snow-cover resolution (VIIRS VNP10A1) | 500 m, daily |
| Climate trend resolution (ERA5) | ~31 km gridded reanalysis |
| Sentinel-2 revisit | ~5 days at mid-latitudes (two-satellite constellation) |
| MODIS/VIIRS archive depth | MODIS from 2000; VIIRS from 2012 |
| ERA5 archive depth | 1940 to near-present |
| Degree-day model lower threshold | ~7°C LST (Ixodes ricinus published literature) |
| Cloud cover limitation | Sentinel-2 optical; MODIS LST gaps filled by 8-day compositing; VIIRS snow masked under dense canopy |
| Output latency (operational update) | Weekly to monthly depending on composite period chosen |
Analytics Satellize can run
| Annual habitat suitability change map | Multi-variable MaxEnt or logistic regression model using LST degree-days, snow-free duration, forest-edge density and NDVI, calibrated against published Ixodes field survey datasets | GIS polygon layer by administrative unit, updated annually, with suitability score and year-on-year delta |
| Seasonal activity-window forecast | Degree-day accumulation from MODIS MOD11A2 above 7°C threshold, with snow-clearance date from VIIRS as season-start anchor | Per-pixel raster of predicted nymphal activity onset and cessation dates, with uncertainty range, delivered before each spring season |
| Expansion front delineation | Threshold exceedance mapping: pixels crossing minimum habitat suitability criteria for the first time in the satellite record, using ERA5 winter minima to confirm thermal barrier breach | Annual boundary shapefile showing newly suitable areas, ranked by population exposure |
| Vaccination prioritisation ranking | Spatial overlay of habitat suitability gain with subnational population and, where available, existing vaccination coverage data | Ranked district table with suitability score, population at new risk, and recommended campaign timing window |
| Multi-year climate trend summary | Linear regression on ERA5 minimum winter temperature and VIIRS snow-free season length per administrative unit over the available archive | Tabular and mapped trend report showing rate of change in key habitat parameters, suitable for inclusion in national health authority briefings |
Who does the work
We can get this done for you. Satellize runs its own analyst desk and a strong science team. You do not buy a data feed and work out what it means; our people source the imagery, run the analysis described on this page, and hand you the answer with its confidence limits stated. Discuss this requirement.