Federal Office of Meteorology and Climatology MeteoSwiss, Zürich, Switzerland.
Int J Biometeorol. 2012 Sep;56(5):949-58. doi: 10.1007/s00484-011-0505-7. Epub 2011 Nov 17.
Detailed knowledge of the spatial distribution of sources is a crucial prerequisite for the application of pollen dispersion models such as, for example, COSMO-ART (COnsortium for Small-scale MOdeling-Aerosols and Reactive Trace gases). However, this input is not available for the allergy-relevant species such as hazel, alder, birch, grass or ragweed. Hence, plant distribution datasets need to be derived from suitable sources. We present an approach to produce such a dataset from existing sources using birch as an example. The basic idea is to construct a birch dataset using a region with good data coverage for calibration and then to extrapolate this relationship to a larger area by using land use classes. We use the Swiss forest inventory (1 km resolution) in combination with a 74-category land use dataset that covers the non-forested areas of Switzerland as well (resolution 100 m). Then we assign birch density categories of 0%, 0.1%, 0.5% and 2.5% to each of the 74 land use categories. The combination of this derived dataset with the birch distribution from the forest inventory yields a fairly accurate birch distribution encompassing entire Switzerland. The land use categories of the Global Land Cover 2000 (GLC2000; Global Land Cover 2000 database, 2003, European Commission, Joint Research Centre; resolution 1 km) are then calibrated with the Swiss dataset in order to derive a Europe-wide birch distribution dataset and aggregated onto the 7 km COSMO-ART grid. This procedure thus assumes that a certain GLC2000 land use category has the same birch density wherever it may occur in Europe. In order to reduce the strict application of this crucial assumption, the birch density distribution as obtained from the previous steps is weighted using the mean Seasonal Pollen Index (SPI; yearly sums of daily pollen concentrations). For future improvement, region-specific birch densities for the GLC2000 categories could be integrated into the mapping procedure.
详细了解源的空间分布是应用花粉扩散模型（例如 COSMO-ART（Consortium for Small-scale MOdeling-Aerosols and Reactive Trace gases））的关键前提条件。然而，对于榛树、桤木、桦树、草或豚草等与过敏相关的物种，并没有这种输入。因此，需要从合适的来源中获取植物分布数据集。我们提出了一种从现有来源中生成此类数据集的方法，以桦树为例。基本思想是使用具有良好数据覆盖范围的区域构建桦树数据集，然后使用土地利用类别将这种关系外推到更大的区域。我们使用瑞士森林清查（1 公里分辨率）与涵盖瑞士非森林区域的 74 类土地利用数据集相结合（分辨率 100 米）。然后，我们将桦树密度类别 0%、0.1%、0.5%和 2.5%分配给 74 个土地利用类别中的每一个。将由此派生的数据集与森林清查中的桦树分布相结合，生成了一个相当准确的涵盖整个瑞士的桦树分布。全球土地覆盖 2000 年（GLC2000；全球土地覆盖 2000 年数据库，2003 年，欧盟委员会，联合研究中心；分辨率 1 公里）的土地利用类别然后与瑞士数据集进行校准，以生成一个覆盖整个欧洲的桦树分布数据集，并汇总到 7 公里 COSMO-ART 网格上。因此，该程序假定在整个欧洲，某个 GLC2000 土地利用类别在任何地方都具有相同的桦树密度。为了减少对这一关键假设的严格应用，可以使用从前面步骤中获得的桦树密度分布来对其进行加权，方法是使用平均季节花粉指数（SPI；每日花粉浓度的年总和）。为了进一步改进，可以将特定地区的 GLC2000 类别桦树密度纳入映射过程。
