Cellular automaton for simulating pollen transport

Some time ago I constructed a cellular automaton (see wikipedia for explanation) for simulating birch pollen concentration in air during the flowering season in the southern Finland. After the serious work was finished, I had some fun re-writing the simulator using Processing, with some simplifications on the physics. This resulted in an applet called SneeFore.

The original CA-simulator was used for educative purposes to illustrate the concentration of airborne birch pollen within the urban areas of Helsinki and Turku cities under similar environmental conditions. The purpose was not to produce a realistic forecasting software, but to introduce an approach which could be used to assess the characteristics of a given region from the perspective of people allergic to birch pollen. The CORINE Land Cover 2000 dataset was used to find the release locations for airborne pollen (see the sidenote below). The atmospheric conditions – wind field, release rate of the pollen and the influence of urban areas – were adjusted to produce qualitatively comparable results for both of the cities.

The original images of the birch pollen concentrations within the Turku and Helsinki areas are below (click for a larger view). In these images the red cover over the Google Map indicates the concentration of airborne birch pollen in inhalation air. The color scale begins from the limit of high concentration value (over 100 grains/m3) in order to reveal the high-end of the concentration spectrum. Thus in practice: when there is red, people will sneeze… Altough the difference between the images of the two cities is rather striking, it mainly boils down to the different forest inventories of the two regions.




The Sneefore Applet

The Processing-based simulator only implements an approximation of the true physics of pollen transport, and has a very restricted geographical area in which you may play in (located in Espoo), but feel free to try the applet below. You may adjust the wind speed, view the CLC2000 land cover raster of the area, set sampling points for displaying the numerical concentration data as a function of time (in logarithmic units), and zoom the pollen raster over Google Maps.
See further instructions by pointing the mouse over the Info icon on the top-right corner of the applet. Go to the Applet.

The pollen release locations were determined using the CORINE Land Cover 2000 -dataset, which contains the land use and land cover classification data for Finland – take a look at the brochure in english. The project for constructing the dataset was co-funded by the European Union and Finland (nationally by The Finnish Environment Institute SYKE, the Ministry of the Environment and the Ministry of Agriculture and Forestry). In practice, the dataset contains a huge GeoTIFF image covering the whole Finland in 25 m resolution:


Published by


Flâneuring through experiments in data, science and artsy digital things.