87
financed by the European Space Agency (ESA). This Land Cover Data will be made
publicly available and, when combined with other statistical data, has the potential
to be used for landscape analysis and for planning of nature conservation actions.
The different types of forest are well understood, as are built-up areas, so the development part of the collaboration entails updates on methodology for large-scale
classifications of open land into varied field layers, starting with the mountain range
and moving on to the rest of the country (Metria 2017; Metria and SLU 2017). The
work will be done during 2017–2019 and includes collating existing data from the
inventories, and creating new data for any vegetation type that is not captured well
enough in the samples (for example seminatural grasslands or some of the more
uncommon types of vegetation in the mountainous areas). The method will encompass a modified version of modelling of desired vegetation types, using Sentinel-2
data, surface layers and wetness indices from; LIDAR data. The sampling design
will then be modified to better capture the vegetation types, both inside and between
the existing NILS squares (Svensson et al. 2017; Reese et al. 2011). The available
infrastructure of annual field workers in NILS enables relatively easy collection of
ground truth-data, and the integration of stereo interpretation ensures that any new
plots are suitable for remote sensing purposes (e.g. being sufficiently large and
homogenous). The long-term aim is to build a database of reference data, which is
accessible to both researchers and authorities for classification purposes.
Challenges in Using Remote Sensing for Monitoring
In any national inventory program, a number of challenges arise. Sweden is an elongated country with a diversity of ecosystems and biogeographical regions, therefore
those interpreting the remote sensing data need a wide breadth of knowledge. This
is often acquired by living in a region or by developing an interest in particular ecosystems and/or how these are used or can be best managed for differing gains.
Recruiting people with sufficient knowledge can sometimes prove difficult.
Whilst remote sensing data provide a valuable source of information for monitoring landscapes, the technologies are constantly evolving: this can compromise
the consistency of observations generally required by monitoring systems. In some
cases, changes may be the result of a difference in observation modes and time
periods. Furthermore, some datasets have only been acquired on relatively few
occasions (e.g., the national LIDAR survey) and hence the lack of repeat coverage
may limit change detection. Alternative methods for retrieving biophysical attributes (e.g., image matching based on aerial photography) may be used but errors in
retrieval are often introduced (Granholm et al. 2015, 2017). Monitoring programs,
therefore, have to be consistent in terms of the data used and knowledge available,
as well as flexible in response to changing technologies and ideas.
NILS – A Nationwide Inventory Program for Monitoring the Conditions and Changes…
financed by the European Space Agency (ESA). This Land Cover Data will be made
publicly available and, when combined with other statistical data, has the potential
to be used for landscape analysis and for planning of nature conservation actions.
The different types of forest are well understood, as are built-up areas, so the development part of the collaboration entails updates on methodology for large-scale
classifications of open land into varied field layers, starting with the mountain range
and moving on to the rest of the country (Metria 2017; Metria and SLU 2017). The
work will be done during 2017–2019 and includes collating existing data from the
inventories, and creating new data for any vegetation type that is not captured well
enough in the samples (for example seminatural grasslands or some of the more
uncommon types of vegetation in the mountainous areas). The method will encompass a modified version of modelling of desired vegetation types, using Sentinel-2
data, surface layers and wetness indices from; LIDAR data. The sampling design
will then be modified to better capture the vegetation types, both inside and between
the existing NILS squares (Svensson et al. 2017; Reese et al. 2011). The available
infrastructure of annual field workers in NILS enables relatively easy collection of
ground truth-data, and the integration of stereo interpretation ensures that any new
plots are suitable for remote sensing purposes (e.g. being sufficiently large and
homogenous). The long-term aim is to build a database of reference data, which is
accessible to both researchers and authorities for classification purposes.
Challenges in Using Remote Sensing for Monitoring
In any national inventory program, a number of challenges arise. Sweden is an elongated country with a diversity of ecosystems and biogeographical regions, therefore
those interpreting the remote sensing data need a wide breadth of knowledge. This
is often acquired by living in a region or by developing an interest in particular ecosystems and/or how these are used or can be best managed for differing gains.
Recruiting people with sufficient knowledge can sometimes prove difficult.
Whilst remote sensing data provide a valuable source of information for monitoring landscapes, the technologies are constantly evolving: this can compromise
the consistency of observations generally required by monitoring systems. In some
cases, changes may be the result of a difference in observation modes and time
periods. Furthermore, some datasets have only been acquired on relatively few
occasions (e.g., the national LIDAR survey) and hence the lack of repeat coverage
may limit change detection. Alternative methods for retrieving biophysical attributes (e.g., image matching based on aerial photography) may be used but errors in
retrieval are often introduced (Granholm et al. 2015, 2017). Monitoring programs,
therefore, have to be consistent in terms of the data used and knowledge available,
as well as flexible in response to changing technologies and ideas.
NILS – A Nationwide Inventory Program for Monitoring the Conditions and Changes…