|
|
Project descriptions
|
|
|
|
English: These webpages are meant for the Scandinavian market, but we have reports in English that may interest you. Send us an e-mail and we will respond!
Below you will find some Norwegian and some English reports. Images say more than the report! |
|
|
|
|
|
|
|
We present here some projects that show what is possible with image analysis. Contact TerraNor for more info. |
|
|
|
|
|
|
Gremeniella attacks |
 |
|
|
|
The Gremeniella project for the Swedish company Stora Enso, we showed that it is possible to map individual trees,
determine tree species and which trees were sick or dead. The project was run in connection with a major greminal attack. |
|
|
|
|
|
|
Measurement of decreases in Kiruna |
|
|
|
|
A new ore has been found in the mines in Kiruna, Sweden. The ore goes under the city. When excavated, the city sinks and the entire city must actually be moved. In this process, LKAB measures the lowering in several ways, including radar satellite. TerraNor was given the task of placing reflectors that help with accurate measurements in given locations.. |
|
|
|
|
|
|
Detect dead and affected trees with ADS 80 sensor from Leica. A project for Swedish Skogsstyrelsen |
This report must be evaluated with the report with the study of Leica ADS 80 sensor images below. They covered the same area and the result was very similar. Both GeoEye and Leica ADS 80 sensor are excellent scanners and well suited for forest mapping.
This report mention briefly the difference between line scanners or push broom scanners and frame cameras. We need to thank
Pierre Labrecque and Alain Gingrais, owners of iCtrees that really did some very good studies using ADS 80. TerraNor learned a lot from them
and their consultant Michel Guerin.
Full article here
|
|
|
|
|
|
|
How to detect Ips typographus L., Spruce bark beetle on GeoEye imagery |
This is a summary in English from a presentation held at EUSI international meeting in June 2013. It must be read together with the report above using Leica ADS 80 sensor.
Michel Guerin did the ITC work using PCI Geoamatica and software from Ph.D Francois Gougeon at the Canadian Forest Service. For more scientific articles, please google "Francois Gougeon".
Full article here
|
|
|
|
|
|
|
Forest mapping with eCognition Developer |
TerraNor has created a ruleset in eCognition which divides the forest into forest stands and then into single trees. Course participants go through each step of the process that can be run fully automatically. Here we describe the process.
Full article here
|
|
|
|
|
|
|
Requirements for the quality of digital data for analysis |
After doing several projects using digital images in Swedish forestry, we sat down with Lantmäteriet (LM) and Stora Enso to discuss what Swedish forestry needs for digital images, or more correctly: what requirements should Swedish forestry place for digital supplies from LM. LM asked for proposals for deliveries and production lines. LM later included most of our proposals in its production line. If you need to order pictures from LM, EUSI or other private actors, this document can give you an overview of concepts that are important.
Full article here
|
|
|
|
|
|
|
|
|
|
|
TerraNor sells satellite images to Swedish Sveaskog |
During the storms Hilde and Ivar (2013/2014), a lot of forest fell in Sweden. This press release shows a photo we took with World-View 2 to find the windfall forest.
Full article here
|
|
|
|
|
|
|
|
ITC, a new and modern method of forest mapping |
|
|
|
This is ITC developed in eCogntion:
Full article here
|
|
|
|
TerraNor offer complete forest survey/mapping
based on new techniques of analysis
of digital image data. With high-resolution data from satellite or aircraft sensors,
we can do classification of tree species, division into stock, calculation of
tree species distribution, volume, forest damage and crown cover.
Important: A good result requires good digital data. TerraNor provides
advice on what data can be used for different types of mapping. A 'pretty picture' is
not necessarily the same as good data. Ask us before you spend money on the sensor
data (digital images), it can pay off. |
|
|
|
'Individual Tree Crown classification' (ITC) is a forest mapping method that divides the forest into single trees. After the classification, classification is run and
determination of tree species for each tree. With all the trees determined by size and tree species, you share
forest into stock. See the photo series below. In the menu of
On the left you can select technical articles and look at examples.
|
|
|
|
The starting point is a high-resolution digital image. Both satellite and airborne sensors
can be used. This is a satellite scene taken with QuickBird (Maridalen May 2003).
The resolution is 0.61m in PAN (black and white) and 2.4m in MS (color). |
|
|
|
Before the division we mask out all areas that are not forests. This makes it easier
to make a good classification. At the same time, customers get updated maps with water, marsh,
roads, plots etc |
|
|
|
The next step is the division of the forest into individual trees. ITC uses a technique
with analysis of the shadows around the trees to determine each tree. See documentation
in the menu on the left. |
|
|
|
To be able to divide the trees into correct classes, we need sample surfaces from fields.
In the test areas, we determine tree species with distribution within the surface. operator
uses the information further in the classification. |
|
|
|
Using the test surfaces, the forest is classified. Which classes to map,
determined by the forest owner. In the Nordic countries there are the usual spruce, pine and pine. In Canada
it is common to classify with 4-5 different hardwoods and 4-5 different softwoods. IN
We have also included this example with older sick spruce forests. Forest damage such as drought
and fungis, are relatively easy to remove if visible in the infrared channels. |
|
|
|
The stock is subdivided at the request of the forest owner. Usually wood species and trees are used
size / age as a classification criterion. In addition, we can use factors such as height
above sea level, steepness, quality, direction, geology etc. The size of the stock may vary
0.5 ha and above.
|
|
|
|
The analysis results in one database with information on each stock.
Databasen contains this info:
|
Area_ha: |
polygon area (ha) |
|
Area_sq_m: |
polygon area (m2) |
|
ITCS: |
total number of ITC crowns |
|
Closure: |
total % of crown cover |
|
Density: |
total stem density (stems / ha) |
|
Sp01: |
tree species # 1 |
|
ITCs01: |
total number of ITC tree crowns for tree species # 1 |
|
Per01: |
% of specie type 1 crowns out of total ITC crowns |
|
CC01: |
% crown cover for specie 1 |
|
Sd01: |
stem density for specie 1 (stems/ha) |
|
Dia01: |
middle crown diameter for specie 1 |
|
Sp02: |
specie # 2 |
|
ITCs02: |
total number of ITC tree crowns for tree species nr 2 |
|
Per02: |
% of specie type 2 crowns out of total ITC crowns |
|
etc…
|
|
Together with tree heights available from Laser / Lidar or statistical field measurements /
image measurements, one can calculate the volume of the different species of wood in the stock.
|
|
|
|
|
|