99
Methods
Data Collection and Pre-processing of Earth Observation
(EO) Data
A wide variety of remotely sensed data have provided observations of the earth’s
environments, with the availability of satellite images increasing exponentially in
recent years. It is, however, very important that datasets are selected based on the
capability of the data to perform the task at hand. This section describes the satellite
and airborne data used in this study.
Very High Resolution Satellite Imagery
One of the many advantages of using satellite imagery for mapping and monitoring
is the possibility of acquiring data regularly with identical sensor specifications.
This increases the potential use, particularly within vegetation monitoring systems,
as the repeatability of surveys is often deemed problematic and renders many methods of data acquisition unsuitable. With a range of Very High Resolution (VHR)
satellites (<2 m pixel resolution) now available, the level of detail seen from space
provides greater opportunities to map and monitor habitats at finer scales.
Images gathered by Worldview-2 are often used to provide detailed classifications of landscapes. This satellite was launched in 2009 and observes in 8 spectral
bands at a spatial resolution of 2 m: a 0.46 m panchromatic band is also available.
In addition to the red, green, blue and Near Infra-Red (NIR) bands, there are four
additional bands that have been created to support specific applications. The coastal
(400–450 nm), yellow (585–625 nm), red edge (705–745 nm) and NIR2 bands
(860–1040 nm) all support vegetation identification and analysis. The red edge band
is particularly important for the analysis of vegetation condition, as changes in chlorophyll production in this region can indicate plant health. Furthermore, Worldview-2
can revisit any site location within one day and is capable of imaging 975,000 km
2
of the land surface on a daily basis (Digital Globe 2009).
As a commercial satellite, Worldview-2 allows users to task image acquisition by
setting the area of interest (minimum of 10 km
2)
and the time window for image
capture. The minimum recommendation for time window duration is 6–8 weeks,
which increases the likelihood of a cloud free acquisition. This amount of control
allows users to capture seasonal variability within the landscape, which is important
for vegetation monitoring and can be used to map habitats that are only spectrally
unique at certain times of the year. For European environments, Lucas et al. (2015)
suggested to use (as a minimum) imagery acquired during the pre- and peak-flush
periods, where the vegetation is relatively stable for extended periods (e.g., no or
full leaf cover). However, additional discrimination can be provided in the transitions from the pre- to the peak-flush and the peak to the post-flush period. The latter
period can be particularly useful for discriminating different plant species. It should
Mapping Coastal Habitats in Wales
Methods
Data Collection and Pre-processing of Earth Observation
(EO) Data
A wide variety of remotely sensed data have provided observations of the earth’s
environments, with the availability of satellite images increasing exponentially in
recent years. It is, however, very important that datasets are selected based on the
capability of the data to perform the task at hand. This section describes the satellite
and airborne data used in this study.
Very High Resolution Satellite Imagery
One of the many advantages of using satellite imagery for mapping and monitoring
is the possibility of acquiring data regularly with identical sensor specifications.
This increases the potential use, particularly within vegetation monitoring systems,
as the repeatability of surveys is often deemed problematic and renders many methods of data acquisition unsuitable. With a range of Very High Resolution (VHR)
satellites (<2 m pixel resolution) now available, the level of detail seen from space
provides greater opportunities to map and monitor habitats at finer scales.
Images gathered by Worldview-2 are often used to provide detailed classifications of landscapes. This satellite was launched in 2009 and observes in 8 spectral
bands at a spatial resolution of 2 m: a 0.46 m panchromatic band is also available.
In addition to the red, green, blue and Near Infra-Red (NIR) bands, there are four
additional bands that have been created to support specific applications. The coastal
(400–450 nm), yellow (585–625 nm), red edge (705–745 nm) and NIR2 bands
(860–1040 nm) all support vegetation identification and analysis. The red edge band
is particularly important for the analysis of vegetation condition, as changes in chlorophyll production in this region can indicate plant health. Furthermore, Worldview-2
can revisit any site location within one day and is capable of imaging 975,000 km
2
of the land surface on a daily basis (Digital Globe 2009).
As a commercial satellite, Worldview-2 allows users to task image acquisition by
setting the area of interest (minimum of 10 km
2)
and the time window for image
capture. The minimum recommendation for time window duration is 6–8 weeks,
which increases the likelihood of a cloud free acquisition. This amount of control
allows users to capture seasonal variability within the landscape, which is important
for vegetation monitoring and can be used to map habitats that are only spectrally
unique at certain times of the year. For European environments, Lucas et al. (2015)
suggested to use (as a minimum) imagery acquired during the pre- and peak-flush
periods, where the vegetation is relatively stable for extended periods (e.g., no or
full leaf cover). However, additional discrimination can be provided in the transitions from the pre- to the peak-flush and the peak to the post-flush period. The latter
period can be particularly useful for discriminating different plant species. It should
Mapping Coastal Habitats in Wales