To Sneeze or Not to Sneeze

When do Allergy Seasons Begin?

It’s August, it’s hot and your dog is itchy.

After ruling out other causes, you’ve done an allergy test and the results are that Brian (your hypothetical dog) has come up to various grasses and weeds, and you want to know how much longer this will go on for.

This leads to some interesting questions. When does a particular allergy season begin and end? And when does it peak? Is it on a set date every year? Perhaps akin to that scene from Father Ted? [1]

Ted: “Any idea why July the 19th should be so important?”

Dougal: “Would that be the day the Ice Age ended?”

Ted: “No, Dougal! They can’t be that precise about the Ice Age!”

That doesn’t feel like the right answer, as we all know how varied the seasons can be. Perhaps it’s a particular month each year? But that’s still not accurate enough. Instead, the start of a given allergy season (technically, the pollinosis season, or the season during which allergic reactions occur) is defined as being when there is a sufficient amount of grass pollen in the air to start causing symptoms in humans.

Studies have shown that the minimum level of airborne grass pollen grains that can lead to such symptoms is 10-50 grains m-3. For example, in one early study from 1973 that measured pollen levels in Cardiff, it was shown that 10% of hay fever patients showed symptoms when grass pollen concentrations exceeded 10 grains m-3 of air [2]. In another study, a level of 50 grains m-3 in London caused all hay fever patients to experience symptoms [3].

The production of pollen is influenced by a large number of factors, classed as primary, secondary and tertiary [4]. These classes refer to the different stages of pollen release, with primary referring to those factors that influence the initial growth of the vegetation (temperature), secondary to those that influence the release of pollen (sunshine, rainfall and relative humidity), and tertiary to those that distribute the pollen (wind). Other factors that can also affect vegetation growth include soil humidity, diseases, pests, pollutants and nutrients.

Definitions of the start date vary. For example, the Threshold 30 method defines the grass allergy season as the first and last days that the pollen concentration reaches 30 grains m-3. Others use some sort of averaging [5], such as the Sum 75 method, which defines the onset as when the cumulated sum for the daily average concentrations reaches 75 grains m-3 (another, the Sum 100 method, uses a value of 100 grains m-3).

There are also definitions that are designed to be applied retrospectively, which allows the duration of the season to be taken into account. For example, the 98% method classes the start of the season to be when 1% of the annual total has been recorded and ends when 99% has been recorded [6]. Similarly, a method by Lejoly-Gabriel defines onset as beginning on

the day when the sum of daily pollen concentrations reaches 5% of the annual total, and when the day in question contributes at least 1%.

Predicting the Seasons

Given these factors, it’s possible to make short-term (a few days) to long-term (an entire pollen season) predictions on pollen levels, including start dates and peak season dates. Such forecasts themselves are often focused on one of three parts of the season; pre-peak, peak, and post-peak.

In a 2005 paper [7], 5 separate mathematical models were used to make medium-range forecasts (~ 7 days) for grass pollen levels in the north of London. These models predicted the peak start date, daily pre-peak levels, daily peak levels, daily post-peak levels, and peak end date, respectively. The researchers used the threshold 30 method mentioned above to designate the start and end of the grass season.

They state that modelling pollen counts in such an urban area presents unique challenges, due to the presence of “high buildings and complex surfaces”, all of which increase turbulence and create considerable variations in pollen counts over relatively short distances. Cities also tend to possess their own microclimates known as ‘urban heat islands’, which promote the upward movement of air and can lead to plants flowering earlier than in the surrounding rural areas.

Models were created by linking previous grass pollen counts with several environmental and meteorological parameters, including 10-day mean maximum and minimum temperatures, windspeed, relative humidity and mean rainfall levels. Their results were modest (62% accuracy when tested against data from 2000), but the links between various factors and pollen seem clear.

Another significant parameter used was something called the North Atlantic Oscillation (NAO) [8], which is a phenomenon relating to fluctuations in atmospheric pressure (switching between the ‘Icelandic low’ and the ‘Azores high’). Crucially, there are known links between the NAO and consequent temperature and precipitation levels in Europe.

Further, in a 2009 paper [9], the links between the NAO and both grass pollen timings and magnitude across Europe were detailed. However, they concluded that other large-scale factors, such as the El Nino Southern Oscillation (ENSO), total solar irradiance (TSI), the volcanic dust veil index (DVI) and surface sea temperatures (SST) may all be required to further enhance the understanding of year-to-year variations. The researchers also concluded that the largest factor in seasonal start times across Europe was latitude, with more northerly regions beginning later.

There is an obvious intermediate step between linking the weather to pollen counts, namely the link between the weather and plant growth. The study of phenology (the study of the timing of biological events such as plant flowering and mammal migration) is widespread in the literature of pollen prediction. For example, a 2015 paper [10] looked at the roles of environmental variables on flowering phenology with an aim to then predict the start of the pollen season for different spring-flowering trees. They found that ‘chilling units’ and ‘forcing units’ (metrics of a plants exposure to chilling and warming temperatures, respectively) were the main factors. Other papers follow a similar theme, focussing on models for ‘budburst’ (the emergence of new leaves after winter hibernation).

There is even a citizen scientist project called Budburst [11] that asks people to contribute phenological observations. From their website they state that “Budburst citizen scientists work together with research scientists, educators, and horticulturists to answer specific, timely, and critical ecological research questions by making careful observations of the timing of plant life cycle events, also called phenophases” and that “These observations are used to better understand how plant species and ecosystems respond to changes in climate locally, regionally, and nationally”.

What does all this mean for vets and pets?

The timing, duration and magnitude of different pollen seasons has clear implications for allergy sufferers. The latest science driving pollen predictions uses a whole host of different parameters to make short, medium and long-term forecasts, including peak daily levels (apparently ragweed levels are at a minimum at 6am [12]), pollen levels for the coming days for a given region, and start dates for allergy seasons months in advance. Such information can guide everything from dog-walking time to when and where to take a holiday, to what level of medication to stock. There are even studies using the latest satellite imagery to create ‘hay fever maps’ based upon high-resolution scans of tree populations [13].

The author of a paper published in 2013 [14], also linking temperature to the start of pollen seasons, states that “The construction of simple models based on thermal conditions has a practical application in allergology, e.g. to estimate the beginning of the allergy immunotherapy in patients sensitive to tree pollen allergens, especially in the case of preseasonal treatment”.

No-doubt such predictions will increase in accuracy in the coming years, as more and more data are compiled and analysed. Soon enough, we’ll all check the pollen count as often as we do the weather, especially thanks to services such as the Met Office already offering daily pollen forecasts.

Then, finally, we can help answer those all-important questions. Should I take the horse out in the early or late morning? Is a family walking holiday in South Wales in July a terrible idea? And just how much will Brian be itching a week on Tuesday?

Written by Rob Harrand – Technology & Data Science Lead


To register your email to receive further updates from Avacta specific to your areas of interest, click here.


  1. Father Ted (‘Hell’ episode)
  2. Hyde, H. A. (1972). Atmospheric pollen and spores in relation to allergy. I. Clinical & Experimental Allergy, 2(2), 153-179.
  3. Davies, R. R., & Smith, L. P. (1973). Forecasting the start and severity of the hay fever season. Clinical & Experimental Allergy, 3(3), 263-267.
  4. Jochner, S., Höfler, J., Beck, I., Göttlein, A., Ankerst, D., Traidl-Hoffmann, C., & Menzel, A. (2013). Nutrient status: a missing factor in phenological and pollen research? Journal of Experimental Botany, 64(7), 2081-2092.
  5. Ong, E., Taylor, P., & Knox, R. (2007). Forecasting the onset of the grass pollen season in Melbourne (Australia). Aerobiologia, 13(1), 43-48
  6. Emberlin, J., Jones, S., Bailey, J., Caulton, E., Corden, J., Dubbels, S., … & Russel, R. (1994). Variation in the start of the grass pollen season at selected sites in the United Kingdom 1987–1992. Grana, 33(2), 94-99.
  7. Smith, M., & Emberlin, J. (2005). Constructing a 7‐day ahead forecast model for grass pollen at north London, United Kingdom. Clinical & Experimental Allergy, 35(10),
  8. North Atlantic Oscillation
  9. Smith, M., Emberlin, J., Stach, A., Rantio-Lehtimäki, A., Caulton, E., Thibaudon, M., Sindt, C., Jäger, S., Gehrig, R., Frenguelli, G., Jato, V., Rajo, F., Alcázar, P., & Galán, C. (2009). Influence of the North Atlantic Oscillation on grass pollen counts in Europe. Aerobiologia, 25(4), 321-332.
  10. Siniscalco, C., Caramiello, R., Migliavacca, M., Busetto, L., Mercalli, L., Colombo, R., & Richardson, A. (2014). Models to predict the start of the airborne pollen season. International Journal of Biometeorology, 59(7), 837-848.
  11. Project Budburst
  12. Jones, B., Barnes, C., Portnoy, J., & Hu, F. (2006). Diurnal variation of airborne ragweed pollen in a metropolitan area-an 8 year perspective. Journal of Allergy and Clinical Immunology, 117(2), S29.
  13. Hay fever map of Britain published to help sufferers avoid hotspots
  14. Myszkowska, D. (2014). Predicting tree pollen season start dates using thermal conditions. Aerobiologia, 30(3), 307-321.