User Information – NVK Skog

Introduction
Forest Conservation Value Map (NVK Skog) is a spatially comprehensive modelling of the relative probability of the occurrence of a value core (“värdekärna) in Swedish forests. The aim is to provide a basis for planning nature conservation at the landscape level. The modelling has been carried out during 2021-2023 as part of a project funded by the Swedish Environmental Protection Agency (see Jonsson et al. 2022).
NVK forest consists of a thematic class based on a modelling of four sub-regions in Sweden: North boreal, south boreal, hemiboreal and nemoral regions. It is provided with 100x100 meter resolution, i.e. with a mapping unit of 1 hectare.

Product description
NVK skog is described in more detail in the report Jonsson et al. 2024, where this user information is included as an appendix.

Geographical reference system
Sweref 99 TM

Dataformat
GeoTiff (tif)

Legends
The relative probability is given as a continuous variable with values ranging from 0 to 100.

Actuality
Training data for modelling consists of the Swedish Environmental Protection Agency's database for forest value cores up to 2016, supplemented with field inventories made in 2019 and 2020
Explanatory variables have been downloaded from open data sources during April 2022, (see Table 1 with associated text in Jonsson et al. 2024) but should be interpreted as valid until 2020. 

Geometric accuracy
On a scale of 1 hectare and includes all forest land where the individual hectare area covers at least 50% forest according to the FAO definition of forest land (tree height >5 m and canopy cover >10%)

The estimated relative probability of value core has been assessed using cross-validation and where the accuracy is very high and varies between 0.89 – 0.90 (ROC based AUC) for each sub-region. See Jonsson et al. (2024) for a description of the validation.

Accessibility
NVK forest is available at the Swedish National Data Service (SND) (https://doi.org/10.5878/wa6j-4b84)

Qualitative assessment of accuracy
NVK forest is modelled for all forest land and is not based on individual forest types. It has not been possible to make a complete evaluation of how the quality varies between forest types, but based on dialogue with experts and the model structure, we note the following guidelines:
- NVK Skog expresses the relative probability of high conservation values per hectare area.
- For the dominant forest types and for larger contiguous areas (> 10 hectares), NVK Skog is reliable. For smaller and isolated areas, NVK Skog should be used with some caution. NKV Skog’s main area of use is to identify larger contiguous areas at the local and regional landscape level.
- Forest in specific sites where conservation values are linked to soil conditions (fertility, soil type, groundwater level) rather than to existing forest structure are not specifically estimated by NVK Skog. This includes, for example, land uplift forests, sand pine forests and coniferous lime forests.
- Also for low-growing, sparse forests on low-productivity land, the estimates are likely to have lower accuracy.
- In highly fragmented forests and with a large proportion of edges, the effect of the surrounding landscape is not captured in the same way as for more contiguous forests. These include, for example, elongated narrow forest areas along watercourses, forests on islands, small forests in agricultural landscapes and peri-urban forests. These conditions affect the accuracy of the estimates.

Complementary geographical data
For practical use, NVK Skog can be supplemented with other geographical information.
- National land cover data (NMD) contains information on the occurrence of different forest types (7 classes and divided into not on wetland and on wetland soils) which enables complementary analyses. It should be noted that the NMD class temporarily non-forest is not included in the modelling, hence the relative probability of value core should be considered non-existent.
- Since NMD describes temporarily non-forest (including clearcuts) up to and including 2018, later years of felling should be taken into account. Information about these can be obtained from the Swedish Forest Agency
- For northern Sweden, the information on the occurrence of continuous cover forests (Ahlkrona et al. 2017) is an important complement and of specific importance for areas below the montane forest line.

In addition to these documents, there are several others that can provide additional information. This includes, among other things, historical maps, soil chemistry, soil and bedrock mapping, occurrences of endangered and red-listed species.

References
Ahlkrona E., Giljam, C., Wennberg, S. 2017. Mapping of continuous cover forests in boreal regions. Metria AB on behalf of the Swedish Environmental Protection Agency, Stockholm, Sweden.
Jonsson, B.G., Angelstam, P., Bubnicki, J.W., Mikusinski, G. & Svensson, J. 2022. Better late than never – indicators of the green infrastructure of the forest landscape. Swedish Environmental Protection Agency, report 7063.
Jonsson, B.G., Angelstam, P., Bubnicki, J.W., Mikusinski, G., Svensson, J. & Undin, M. 2024. Forest: A probability model for conservation values on forest land. Swedish Environmental Protection Agency, report under print