94
biogeochemical factors of the ecosystem (Menzel 2002). For example, time-series
NDVI data have been used to indicate changes in LAI globally, with these reflecting
human-induced and natural events and processes, including those related to climactic fluctuation (Liu et al. 2010). Phenology is also important for estimating biological productivity, understanding land-atmosphere interactions and the management
of vegetative resources (Lieth 1971; Taylor 1974; Sarmiento and Monasterio 1983).
Therefore, it is a key factor in mapping to the species level using EO data and also
essential for monitoring change. Additionally, timing the acquisition of remotely
sensed datasets to coincide with critical phenological stages of flowering or leaf
senescence is very important when mapping invasive species (He et al. 2011).
In recent years, Very High Resolution datasets have increased in popularity as
there is the potential to resolve more habitat categories, but the lack of
shortwave-infrared band in these datasets has significantly hampered their potential
for monitoring complex environments. However, sensors such as Worldview-2 and
Worldview-3, with additional coastal, yellow, red edge and near infrared bands are
anticipated to provide benefits over other VHR sensors such as IKONOS, Quickbird
and GeoEye. Techniques and software for processing these data, in addition to SAR
and LiDAR data, are also likely to become more available in future years, and the
increase in open source material will benefit many managers of protected areas in
countries where funding is more limited (Nagendra et al. 2013). However, it is recognised that more effort should be put into developing a coherent and operational
method which produces most, or all, relevant parameters to contribute to assessment
of conservation status (Corbane et al. 2015).
It is also important to have well designed programmes of field data collection
that maximise the use of data for remote sensing interpretation and conservation
assessments. In situ field sampling networks therefore, need to be designed in combination with remote sensing using, for instance, stratified sampling designs to carefully assess species distributions across different habitat types and enhance
interpretative power (Nagendra 2001). Unless the spatial grain, extent and timing of
remote sensing data, in situ data and models are well matched to these features of
the habitats, the robustness of conclusions on management effectiveness, and the
interpretive power of the analytical techniques used, will be limited. Remote sensing interpretation needs to be grounded in field data, as this is critical for effective
adaptive management and monitoring (Nagendra et al. 2013).
Image Analysis Techniques
Image analysis within the remote sensing community traditionally refers to image
classification, which is described as the systematic grouping of classes or themes
extracted from remotely sensed data: it is a preferred technique because the methods
are well known and widely used within the community. The output is generally
G. Jones et al.
biogeochemical factors of the ecosystem (Menzel 2002). For example, time-series
NDVI data have been used to indicate changes in LAI globally, with these reflecting
human-induced and natural events and processes, including those related to climactic fluctuation (Liu et al. 2010). Phenology is also important for estimating biological productivity, understanding land-atmosphere interactions and the management
of vegetative resources (Lieth 1971; Taylor 1974; Sarmiento and Monasterio 1983).
Therefore, it is a key factor in mapping to the species level using EO data and also
essential for monitoring change. Additionally, timing the acquisition of remotely
sensed datasets to coincide with critical phenological stages of flowering or leaf
senescence is very important when mapping invasive species (He et al. 2011).
In recent years, Very High Resolution datasets have increased in popularity as
there is the potential to resolve more habitat categories, but the lack of
shortwave-infrared band in these datasets has significantly hampered their potential
for monitoring complex environments. However, sensors such as Worldview-2 and
Worldview-3, with additional coastal, yellow, red edge and near infrared bands are
anticipated to provide benefits over other VHR sensors such as IKONOS, Quickbird
and GeoEye. Techniques and software for processing these data, in addition to SAR
and LiDAR data, are also likely to become more available in future years, and the
increase in open source material will benefit many managers of protected areas in
countries where funding is more limited (Nagendra et al. 2013). However, it is recognised that more effort should be put into developing a coherent and operational
method which produces most, or all, relevant parameters to contribute to assessment
of conservation status (Corbane et al. 2015).
It is also important to have well designed programmes of field data collection
that maximise the use of data for remote sensing interpretation and conservation
assessments. In situ field sampling networks therefore, need to be designed in combination with remote sensing using, for instance, stratified sampling designs to carefully assess species distributions across different habitat types and enhance
interpretative power (Nagendra 2001). Unless the spatial grain, extent and timing of
remote sensing data, in situ data and models are well matched to these features of
the habitats, the robustness of conclusions on management effectiveness, and the
interpretive power of the analytical techniques used, will be limited. Remote sensing interpretation needs to be grounded in field data, as this is critical for effective
adaptive management and monitoring (Nagendra et al. 2013).
Image Analysis Techniques
Image analysis within the remote sensing community traditionally refers to image
classification, which is described as the systematic grouping of classes or themes
extracted from remotely sensed data: it is a preferred technique because the methods
are well known and widely used within the community. The output is generally
G. Jones et al.