The Science Behind Our Pollen Forecasts

Pollen forecasting isn't guesswork. It's atmospheric physics, plant biology, and decades of observational data, stitched together into a system that tells you what's in the air before you step outside. Here's how the Atmospore model works, why it works at the resolution it does, and what the peer-reviewed literature says about predicting airborne allergens.

Why Our Resolution Hits the Sweet Spot

Our model runs at a resolution of roughly 25 km between grid points. That might sound coarse compared to, say, a street-level air quality sensor. But pollen doesn't work like car exhaust. Airborne pollen concentrations are governed by mesoscale weather: wind fields, frontal passages, precipitation, and temperature gradients that operate over tens of kilometres. Our grid captures exactly these drivers.

This isn't just our claim. In a peer-reviewed study,¹ researchers ran an atmospheric transport model at 36 km resolution across the entire contiguous United States and found it successfully reproduced observed seasonal pollen distributions (Pearson r = 0.35–0.40 against 58 monitoring stations). Sub-kilometre variation does exist within cities. One study² documented up to 300% differences across Berlin. But that variation comes from individual trees and local wind channelling, not from anything a forecast can predict. Our resolution captures what's forecastable. Everything finer would be noise dressed up as signal.

How Pollen Moves Through the Atmosphere

A birch pollen grain is roughly 22 micrometres in diameter and weighs around 5–10 nanograms. Once released from the catkin, it behaves like a fine aerosol particle. Wind carries it horizontally while gravitational settling and turbulent diffusion control its vertical profile. On a dry, windy spring morning, birch pollen can reach 2 km altitude and travel 100+ km before depositing. Wet deposition (rain washing pollen from the air) is the primary removal mechanism and can cut concentrations by 80% within hours.³

Species matter too. Most people are allergic to specific species, not to pollen in general. Someone who reacts strongly to birch may tolerate oak without symptoms. On top of that, individual pollen grains vary in how much allergen they carry. Research has shown that birch pollen from different regions can contain very different levels of the Bet v 1 protein that triggers the immune response.⁵ A total pollen count doesn't tell you much if you don't know which species are in the air. That's why Atmospore forecasts at the species level.

Our model accounts for a wide range of environmental dynamics using established physics. The result is a reliable model-grounded forecast, not a statistical interpolation between monitoring stations.

A Multi-Stage Forecast Pipeline

Producing a species-specific daily pollen number for any land grid point on Earth requires a vast amount of quality input data and computational power. Thanks to our pipeline and setup, we are able to capture and model the specifics of each plant species.

We model the seasonal life-cycle of the species using established phenological principles, producing a daily species-specific pollen concentration at each grid point. That's the number that matters for your nose.

Validated Against Real Observations

We continuously calibrate our models against a wide array of real measurements from across the continents. Each species model is finely tuned with state-of-the-art technology and continuously optimised.

The model is never static. Every season brings new observations, and we continuously improve the models. We model species individually so that the forecast tracks each one's distinct behaviors. The goal is always the same: if you're allergic and it's in the air, we want the forecast to say so.

References

1. Ren, X., Cai, T., Mi, Z. et al. (2022). Modeling past and future spatiotemporal distributions of airborne allergenic pollen across the contiguous United States. Frontiers in Allergy, 3. PMC9640548
2. Werchan, B., Werchan, M., Mucke, H-G. et al. (2017). Spatial distribution of allergenic pollen through a large metropolitan area. Environmental Monitoring and Assessment, 189(169). PMID: 28316024
3. Sofiev, M. et al. (2013). A numerical model of birch pollen emission and dispersion in the atmosphere. Description of the emission module. International Journal of Biometeorology, 57, 45-58.
4. Anderegg, W.R.L. et al. (2021). Anthropogenic climate change is worsening North American pollen seasons. Proceedings of the National Academy of Sciences, 118(7).
5. Buters, J.T.M. et al. (1997). Release of Bet v 1 from birch pollen: variable allergen content per grain. Journal of Allergy and Clinical Immunology. JACI