13
computer- intensive statistical procedures, or citizen science, each of which requiring specialized skills and (often expensive) equipment.
In this paper, we focus on another critical factor that, in our view, may form a
major barrier to the operationalization of remote sensing in nature conservation
applications, but is all too often overlooked. That factor is method transferability, or
more precisely, the lack thereof. We argue that poor transferability of remote sensing methods may be the cause of many disappointments among potential users, and
therefore hinder further attempts to their operational implementation in biodiversity
conservation. We discuss different causes of poor method transferability, and illustrate this with some examples based on Natura 2000 habitat monitoring, an application field where operationalization of remote sensing monitoring could have a large
impact. Finally, we discuss pathways to overcome the current mismatch, and make
remote sensing applications more widely useful and transferable.
Transferability as a Critical Success Factor: The Case
of Natura 2000 Habitat Monitoring
Vanden Borre et al. (2011b) noticed a low uptake of operational remote sensing in
Natura 2000 habitat monitoring. Several possible causes were put forward, both on
the technical and on the practical side. Technical challenges include, for instance, the
fact that habitat types strongly vary in scale, show high variation across Europe, and
sometimes can only be reliably identified by the presence of a selected set of indicator species (the latter mostly occurring in low numbers) (Vanden Borre et al. 2011a).
Challenges of a more practical kind are inter alia the difficulty of obtaining suitable
image data at a suitable time, and the potentially high cost associated with that. Also,
many potential user organizations lack the profound GIS and remote sensing skills
needed to effectively make use of this type of information (Kennedy et al. 2009).
While each of these causes may be valid, another aspect has, until now, been
largely ignored but could prove to be a critical factor, namely: a lack of method
transferability. With the right kind of data and properly fine-tuned methods, remote
sensing can deliver highly detailed spatial and thematic information. Many biodiversity monitoring schemes, such as Natura 2000 habitat monitoring, require, or at
least benefit from, high spatial and thematic detail. However, when details are
important, and the potential to extract such detail is there, approaches not surprisingly tend to be specifically geared to the problem at hand. After all, remote sensing
scientists, like other scientists, need to publish to persist. And failures usually do not
make it to publication (Fanelli 2012; Matosin et al. 2014). Additionally, in a methodfocused field like remote sensing, coming up with novel methods further increases
the chances of getting the work published. In recent years, the number of published
methods for detailed remote sensing of Natura 2000 habitats has soared. Many of
these include extensive fine-tuning to tackle complex problems. As such, the methods are often adjusted to one particular location, to such an extent that it may jeopardize their applicability elsewhere.
Towards a Mature Age of Remote Sensing for Natura 2000 Habitat Conservation: Poor…
computer- intensive statistical procedures, or citizen science, each of which requiring specialized skills and (often expensive) equipment.
In this paper, we focus on another critical factor that, in our view, may form a
major barrier to the operationalization of remote sensing in nature conservation
applications, but is all too often overlooked. That factor is method transferability, or
more precisely, the lack thereof. We argue that poor transferability of remote sensing methods may be the cause of many disappointments among potential users, and
therefore hinder further attempts to their operational implementation in biodiversity
conservation. We discuss different causes of poor method transferability, and illustrate this with some examples based on Natura 2000 habitat monitoring, an application field where operationalization of remote sensing monitoring could have a large
impact. Finally, we discuss pathways to overcome the current mismatch, and make
remote sensing applications more widely useful and transferable.
Transferability as a Critical Success Factor: The Case
of Natura 2000 Habitat Monitoring
Vanden Borre et al. (2011b) noticed a low uptake of operational remote sensing in
Natura 2000 habitat monitoring. Several possible causes were put forward, both on
the technical and on the practical side. Technical challenges include, for instance, the
fact that habitat types strongly vary in scale, show high variation across Europe, and
sometimes can only be reliably identified by the presence of a selected set of indicator species (the latter mostly occurring in low numbers) (Vanden Borre et al. 2011a).
Challenges of a more practical kind are inter alia the difficulty of obtaining suitable
image data at a suitable time, and the potentially high cost associated with that. Also,
many potential user organizations lack the profound GIS and remote sensing skills
needed to effectively make use of this type of information (Kennedy et al. 2009).
While each of these causes may be valid, another aspect has, until now, been
largely ignored but could prove to be a critical factor, namely: a lack of method
transferability. With the right kind of data and properly fine-tuned methods, remote
sensing can deliver highly detailed spatial and thematic information. Many biodiversity monitoring schemes, such as Natura 2000 habitat monitoring, require, or at
least benefit from, high spatial and thematic detail. However, when details are
important, and the potential to extract such detail is there, approaches not surprisingly tend to be specifically geared to the problem at hand. After all, remote sensing
scientists, like other scientists, need to publish to persist. And failures usually do not
make it to publication (Fanelli 2012; Matosin et al. 2014). Additionally, in a methodfocused field like remote sensing, coming up with novel methods further increases
the chances of getting the work published. In recent years, the number of published
methods for detailed remote sensing of Natura 2000 habitats has soared. Many of
these include extensive fine-tuning to tackle complex problems. As such, the methods are often adjusted to one particular location, to such an extent that it may jeopardize their applicability elsewhere.
Towards a Mature Age of Remote Sensing for Natura 2000 Habitat Conservation: Poor…