Tree Species Distribution

Current tree species distributions are mapped using (I) the Forest Focus (2003) database, (II) different point surface interpolation methods and (III) the pan-European forest/non-forest Map (Pekkarinen et al., 2008).

Maps are available for download here.

Tree species distribution maps are useful in several forestry applications. In the context of the impacts from climate change they are useful to compare between current potential and observed distribution of species, and as reference between current distribution and future potential distributions.

Below we present a short summary of the implementation of the species distribution maps at 1km grid size updated for the years 2004/2005. Two point surface interpolation methods are used: Inverse Weighted Distance (IWD) and Nearest Neighbour applied to field samples from the Forest Focus database and subsequently filtered with the Pan European forest/non-forest map year 2000 rescaled to 1km grid size.

METHODOLOGY
1. Data preparation
Geographical locations and relative percentage presence were extracted for each one of the 120 species existing in the Forest Focus databases level I and II. The resulting table (table 1) comprises more than 6600 field locations with an average distance between locations of ~13km. See location of field survey in figure 1.



2. Inverse Weighted Distance interpolation
The surface approximation of point data was carried out using IWD (Watson and Philip, 1985; Phillips et al., 1992). Further, a second point surface interpolation method, Nearest Neighbour, was applied.

3. Species relative presence map
From the IDW interpolation species maps, the relative presence in percentage of each species "S" to the overall sum of 120 species presence is computed. As a result summing the 120 species maps, each pixel is equal to 1.

4. Species distribution maps in ha of tree cover per species

The Pan European forest/ non-forest map was rescaled to a 1km grid size. In the rescaled map the values of the pixels range from 0 to 100 ha tree cover. The rescaled forest map is multiplied with each one of the 120 relative percentage presence specie maps. The final maps estimate the area of tree cover per species in ha within each grid cell. In order to back validate this procedure, the sets of all species distribution maps are summed and the resulting pixel value is equal to the resampled forest/ non-forest map.

RESULTS
Current species distribution maps are available to download and visualize here

SUMMARY
The main result is European species distribution maps at 1km grid size. The implementation of the forest/non-forest map filter allowed producing maps of upmost quality in terms of distribution (where the species growth should reflect the reality of the landscape).

When using the maps, some limitations and mapping errors should be considered:

• Accuracy: how close is a measured value to the actual true value? The forest focus database was not conceived to estimate the proportion of different species relative cover. Nevertheless, the European forest landscape is characterized by a low tree cover differentiation. Mixed forests are mostly composed by two to four species and most of the forests are monospecific. This means the Forest Focus database represents the correct proportion of species differentiation with a sufficient accuracy in most locations.

• Interpolation errors. IDW interpolated values and the resulting accuracy of species surface maps, are a function of the distance from the measured field locations. Within ~13 km average distance between input plots location, IWD describes correctly the terrain reality in homogeneous forest landscapes. When forest management is locally highly differentiated or the terrain is steep with high altitude gradients or abrupt soil difference exists, IWD can under- or over-estimate the proportion of a particular species or exclude/include it from the map. That limitation of the map should be taken into account when using the products.

Further studies using for instance forest inventory databases with higher density of samples and/or using a more sophisticated interpolation algorithm (e.g. a geographically weighted regression), could significantly improve the quality of the species distribution maps.

REFERENCES

• Forest focus (2003). Regulation (EC) no 2152/2003. Forest Focus: monitoring of forests and environmental interactions in the community.
• Pekkarinen, A., Reithmaier, L., Strobl, P. (2009): Pan-European forest/non-forest mapping with Landsat ETM+ and CORINE Land Cover 2000 data. ISPRS Journal of Photogrammetry and Remote Sensing, 64, 171-183.
• Phillips, D., Dolph, J., Marks, D. (1992): A comparison of geostatistical procedures for spatial analysis of precipitation in mountainous terrain. Agricultural and Forest Meteorology, 58, 119-141.
• Watson, D., Philip, G. (1985): A refinement of inverse distance weighted interpolation. Geo-Processing, 2, 315-327.

Mission

As the science and knowledge service of the European Commission, the Joint Research Centre's mission is to support EU policies with independent evidence throughout the whole policy cycle.