Environmental Engineering Reference
In-Depth Information
other species the length of day or ground moisture are the important factors. In the
current work, we consider the forecasting of birch and grass pollen concentrations.
Birch is a tree, which has got light pollen grain and its flowering is well described
by the thermal time type model. Usually, birch flowering lasts a couple of weeks.
Grasses include many different species and thus the flowering period of a
“general grass” lasts for several weeks or even months. In case of grass pollen the
source is near the ground and grains themselves are usually heavier than those of
birch. Thus, grass pollen is rarely transported over very long distances.
In the current work we compare the quality of the forecasts for different type of
pollen grains and analyse the sensitivity of the model performance to its para-
meters: low versus high source assumption, short versus long flowering period,
light versus heavy pollen grain and also an ability of the model to use fixed
starting time of flowering in real-life situations. We will also highlight the weather
parameters, which have the most important role in numerical pollen forecasts.
2. Material and Methods
Pollen long-range transport forecasts are performed with the dispersion modelling
system SILAM (Sofiev et al., 2006b). It is capable of both forward and inverse
simulations, for which Lagrangian random-walk particle model and Eulerian
dynamic cores are available. Current pollen simulations use SILAM v4.2.1 with
Eulerian dynamics. The system is based on input flow of meteorological infor-
mation obtained from numerical weather prediction (NWP) models. Here we have
used ECMWF data with 3 h resolution.
In case of birch, both phenological models for the starting and ending date of
flowering and a pollen release model itself need information from the NWP model.
This information controls also transport with wind, mixing due to turbulence, and
sink processes removing the pollen grains from the air. Release of pollen grains,
as presented in our model, is a complex system, which uses a map of the fraction
of birch forests in European landscape (Sofiev et al., 2006a), and computes the
starting date of flowering using Thermal Time-type phenological model (e.g.,
Häkkinen et al., 1998; Linkosalo, 2000). The emission module also utilizes the
number of catkins per tree (relative to normal for each specific region). Dynamics
of the emission is driven by the pollen release model. More details about the
SILAM pollen forecasting system can be found from Siljamo et al. (2008).
In case of grass, maps of fixed calendar dates for starting and ending the
flowering was computed from grass pollen observations of the European
Aeroallergen Network (EAN). Pollen release model is similar to that of birch.
Birch and grass pollen observations used for the model evaluation were available
from EAN(European Aeroallergen Network) database, which receives the data
from about 35 countries and about 300 sites. Pollen observations began in 1974
but the bulk of data is after 1985.
