National
Forest
Inventory

Executive summary

General information

History

Materials

Methods

Results

Resources

Expertise

Products

Publications

Applications of results

Current events/projects

Use of SAR (radar) data

Imaging spectrometry

Dynamics of biodiversity

[Finnish]
METLA
METLA Research

MULTI-SOURCE NATIONAL FOREST INVENTORY OF FINLAND, MATERIALS

The Field Sampling System

The distance between two tracts varies from south to north, with the density and variableness of forests and is 7 kilometres both in north-south and east-west directions in Northern part of Central Finland. One tract has 15 sample plots, of which three are permanent and the other twelve temporary (Tomppo 1993). Global positioning system (GPS) with real time correction and radio data system (RDS) has been recently intoduced. It allows, in addition to localization of sample plots, a navigation to the plots.

The coordinates of the trees on permanent plots are registered in order to identify them during the next inventory.

A Bitterlich sample plot is applied. Tallied trees are selected with a relascope, the relascopic factor varies by region, from south to north and west to east, depending on the density of the forests and is 1.5 in Central Finland. The maximum radius is 12.45 metres (corresponding to the breast-height -diameter 30.5 cm with the factor 1.5). Reducing the radius of a sample plot decreases the reliability of estimates very little, but decreases in some cases the amount of field work noticeably, because the number of divided sample plots decreases. Every 7th tally tree is measured as a sample tree (Tomppo 1993).

Satellite Image Data

Digital Map Data

Arable land, urban areas, single buildings, roads, swamps and digital terrain model are in digital form, at the moment. Some administrative information such as municiple boundaries, boundaries of areas of ownershipsgroups and, in the future, boundaries of forest holdings will also be used in digital form in order to differentiate computations units. Water areas could be obtained from base maps but they can be obtained relatively reliably from satellite images as well as peat production areas, when they are not available on maps. Digital map data are transformed into raster form, digital map themes corresponding one and boundaries of computation units another channel of the file.

1) Arable land

The information on arable land is based on the base map of the scale 1:50 000. The yellow arable land element has been scanned by National Land Survey of Finland with a pixel size of 25 m by 25 m At the moment, pure map data have been used for the separation of land classes. A combination of a digital analysis of satellite images and digital map information will be used in the future for classifying agricultural areas, because the map data are not necessarily up-to-date.

2) Urban areas and Buildings

The information on built-up land is obtained from the housing register provided by the Civil Register of Finland. The coordinates of each house in Finland are known. This information can be utilized for producing a mask of land occupiad either by urban areas or detached houses.

A group of houses with at least 200 inhabitants and with a within houses distance at most 200 metres form an urban area and are combined into a connected built-up land. These vector form data are purchased from Statistical Centre of Finland and is rasterized with ARC/INFO utilities into a BIL- or SUN -raster file. This raster file is transformed into a one-bit DISIMP -raster file. The urban areas are indicated by the value 1 and are masked into the map -file with the intensity value 4.

Buildings outside build-up areas are presented as squares of a size of 2 x 2 pixels. The four pixels which are closest in the geographical space to the house and which compose a square are applied. The coordinates of houses are received from National Land Survey of Finland with the permission of the Civil Register of Finland.

3) Roads

The road maps are received from National Land Survey of Finland in vector files of DXF -format. The coordidates of corner points of roads are listed in the file. The points of vector file is connected to solid lines using an ARC/INFO utility and transformed into raster-file with an utility written at METLA. At the beginning, all roads were of width of 25 metres. Recently, a new system has been introduced. Roads are classified by means of road class number and the width of each road in the raster file depends on the class, varying from 25 to 100 meters.

4) Digital boundaries of computation units

The computation units in ordinary image processing are municipalities. Digital municipal boundaries have been purchased from the National Land Survey of Finland with a format of FINGIS transfer file. The boundaries have been digitzed from maps with a scale of 1:10000. For image processing, the vector format file are converted into DISIMP -raster format in such a way that the intensity value of a pixel is the same as the index number of that municipality which involves the pixel.

Result computation is needed often by request for some other computation units than municipalities, for instance for different ownershipgroups by municipalities or even for forest holding of a single owner.

The main ownershipgroups in Finland are private owners (63 % from the area of forest land) state, forests managed by Finnish Forest and Park service (24 %) and forest industry companies (9 %).

The task is thus to separate state and companies owned forests from private owned forests. There are several optional methods for this purpose. A part of forest holding boundaries of forests of Forest and Park service are already in a digital form. In other cases, digitazing of boundaries with a coordidate reader or scanning of colour elements of a map which indicates certain ownershipgroups or an ownership (e.g. forest industry company) can be used.

For image processing, administrative units and owner groups are transformed into raster format.

5) The digital terrain model

Digital terrain models are used for correcting original spectral values in order to avoid confusion in image analysis caused by land morphology. A digital elevation model of the entire country has been bought from National Land Survey of Finland. It is of a raster format file with the pixel size of 25 m x 25 m and with the resolution of 10 cm. This file has produced from the original elevation contour of the Finnish base map. The origal countour interval has been either 5 or 2.5 metres.

6) Peatland

The spectral response of peatlands (swamps) differs from that of the mineral soils with the same growing stock. Further, some peatlands can not be separated from mineral soils. Therefore, digital peatland information is used in order to improve the accuracy of estimates.

Data on peatlands has been scanned from basic maps (scale 1:100,000) or GT-maps (scale 1:200,000). The data is provided by the National Land Survey of Finland in raster format. Tests have shown, however, that these data is not always reliable enough to be used in the image processing in a categoric way, i.e. field measuremet based peatland plots are applied with pixels belonging to peatland mask and vice versa. Therefore, peatland mask is utilised in image processing in such a way that sample plots laying on pixels which belong to the peatland mask is applied with those pixels which belong to the area of peatland mask and vice versa. This means that a stratification for both the sample plots and the satellite images is carried out by means of peatland mask (not with field observed peatland class for the sample plots) and the strata are processed independtly.

7) Clouds

In some cases clouds and/or their shadows cover a part of the image to be processed. The total area of the image affected by clouds or their shadows are delineated. A colour composition of visible channnels (TM -channels 1, 2 and 3) of the image are transferred to the display unit of a workstation and enchanged in such a way that also crepelike clouds and their shadows can be visually recognised as well as possble. The areas are surrounded (manually with a mouse) by polygons. These areas are transformed into a 1-bit raster file. This raster file is an optional input file in combining intensity values to ground sample plots and in image processing. The sample plots whose intensities are affected by clouds or shadows are left out from ground truth data. The pixels affected by clouds or shadows are not processed in the image anylysis. The forest area under clouds or shadows is assumed to be similar to the avarage of the rest of the computation unit (e.g. municipality).

8) Water

Water areas are delineated by the information of satellite images at the moment. This information is quite reliable for for the purpose. Channels 4 and 7 of Landsat TM images and tresholding are employed.

9) Peat production areas

Some of the peat production areas are delineated on maps. These areas are digitized and processed as any other pixels on non- forestry land. If data on peatland production areas are not reliable or complete, numeric classification is used. Reference areas are taken from such land that can reliably be classified as peat production area and land classified as forestry land. Reference areas and discriminant analysis based on generalized squared distances are used for classification of peat production areas. DAN program written at FFRI are used for the discrimination.

References

Tomppo, E. 1993. Multi-Source National Forest Inventory of Finland. Proceedings of Ilvessalo Symposium on National Forest Inventories. Finland 17-21 August, 1992. IUFRO S4.02. The Finnish Forest Research Institute. Research Papers 444. pp. 52-60.