83
Inventory by Remote Sensing
Aerial imagery is used in many countries to support environmental mapping and
monitoring, and is discussed in conservation research (e.g. Rose et al. 2014).
Examples include the UK Countryside Survey (Brandt et al. 2002; Barr et al. 2003),
and the Norwegian Monitoring System, 3Q, (Dramstad et al. 2002). As with Sweden,
Estonia used stereo near-infrared imagery to sample the landscape, and initial tests
have been successfully carried out in Scotland for monitoring of Natura 2000- areas
(Roose et al. 2007; Mattisson and Sullivan 2017). The advantage of using stereo
near-infrared imagery is that different land cover (including vegetation) types can
be differentiated and their health also indicated (Ihse 2007). The greatest benefit
arises when these are interpreted by those with knowledge of the landscape and its
ecology. Whilst individual species are often difficult to discriminate using colour
infrared imagery, this can be achieved where they become dominant. However, most
mapping is based on life-forms and their relative distribution in the landscape and
this is generally the basis for mapping land cover, use and change as a consequence
of natural process and human activities.
The military forces first established the benefits of near-infrared imagery as differences in reflectance separated camouflaged vehicles and guns from natural leaf
cover. Since then, it has been used extensively in Sweden. An example is given in
Fig. 2, which compares a colour infrared image with delineated land covers (Allard
2012b). The landscape is a typical setting in the mid-eastern part of Sweden, where
the former rural areas have been mostly abandoned to regrowth. The foreground
shows grazing land (1), with bedrock outcrops and woody shrubs on land that was
never farmed or fertilized and is of comparatively high biological diversity. The area
may still be grazed but the intensity is low. A second area (2) represents an area that
may historically have been used to produce hay but where the unevenness of the
ground hindered ploughing and hence only grazing has taken place. Recent abandonment of the area is indicated by the regenerating deciduous trees and some
elderly birches and oaks. The third area (3) represents a farmed field that has been
abandoned but where the addition of fertilizers has led to differences in the composition of herbaceous (grasses, forbs) and woody shrubs. The final area (4) is a mix
of former grazing land but also mature coniferous forests that has been managed for
timber (house and tool construction). Each of these units can be differentiated
within the colour aerial photography.
Throughout Sweden, changes in the landscape are varied but can be quantified
from the time-series of aerial images available: the nation was fully covered by
stereo images in the 1950s–1960s and again in the 1970s–1980s, which presents
good opportunities to ‘hindcast’ the estimations backwards in time. To date, all of
the sample squares of the 1970s–1980s in farming areas have been fully inventoried: pastures and artificial land types (e.g. built-up areas) and agricultural areas
have been compared (Christensen et al. 2015), and there is an on-going inventory
that goes further back in time into the 1950–1960s. The results of this inventory will
be compared to European efforts (Köhler et al. 2006). As examples of hindcasting,
NILS – A Nationwide Inventory Program for Monitoring the Conditions and Changes…
Inventory by Remote Sensing
Aerial imagery is used in many countries to support environmental mapping and
monitoring, and is discussed in conservation research (e.g. Rose et al. 2014).
Examples include the UK Countryside Survey (Brandt et al. 2002; Barr et al. 2003),
and the Norwegian Monitoring System, 3Q, (Dramstad et al. 2002). As with Sweden,
Estonia used stereo near-infrared imagery to sample the landscape, and initial tests
have been successfully carried out in Scotland for monitoring of Natura 2000- areas
(Roose et al. 2007; Mattisson and Sullivan 2017). The advantage of using stereo
near-infrared imagery is that different land cover (including vegetation) types can
be differentiated and their health also indicated (Ihse 2007). The greatest benefit
arises when these are interpreted by those with knowledge of the landscape and its
ecology. Whilst individual species are often difficult to discriminate using colour
infrared imagery, this can be achieved where they become dominant. However, most
mapping is based on life-forms and their relative distribution in the landscape and
this is generally the basis for mapping land cover, use and change as a consequence
of natural process and human activities.
The military forces first established the benefits of near-infrared imagery as differences in reflectance separated camouflaged vehicles and guns from natural leaf
cover. Since then, it has been used extensively in Sweden. An example is given in
Fig. 2, which compares a colour infrared image with delineated land covers (Allard
2012b). The landscape is a typical setting in the mid-eastern part of Sweden, where
the former rural areas have been mostly abandoned to regrowth. The foreground
shows grazing land (1), with bedrock outcrops and woody shrubs on land that was
never farmed or fertilized and is of comparatively high biological diversity. The area
may still be grazed but the intensity is low. A second area (2) represents an area that
may historically have been used to produce hay but where the unevenness of the
ground hindered ploughing and hence only grazing has taken place. Recent abandonment of the area is indicated by the regenerating deciduous trees and some
elderly birches and oaks. The third area (3) represents a farmed field that has been
abandoned but where the addition of fertilizers has led to differences in the composition of herbaceous (grasses, forbs) and woody shrubs. The final area (4) is a mix
of former grazing land but also mature coniferous forests that has been managed for
timber (house and tool construction). Each of these units can be differentiated
within the colour aerial photography.
Throughout Sweden, changes in the landscape are varied but can be quantified
from the time-series of aerial images available: the nation was fully covered by
stereo images in the 1950s–1960s and again in the 1970s–1980s, which presents
good opportunities to ‘hindcast’ the estimations backwards in time. To date, all of
the sample squares of the 1970s–1980s in farming areas have been fully inventoried: pastures and artificial land types (e.g. built-up areas) and agricultural areas
have been compared (Christensen et al. 2015), and there is an on-going inventory
that goes further back in time into the 1950–1960s. The results of this inventory will
be compared to European efforts (Köhler et al. 2006). As examples of hindcasting,
NILS – A Nationwide Inventory Program for Monitoring the Conditions and Changes…