148
Field Data
Field data is needed from the study area describing the plant communities present
as auxiliary information for planning the airborne campaign, supporting the imagery analysis, and validating the cartography generated by the process. The user
organization plays a key role in the collection and preparation of these field data.
(i) The topographic and plant community maps of the study area are used to plan
the hyperspectral flights and provide cover estimates of the plant species. A
Digital Elevation Model (DEM) with high spatial resolution is needed to support the processing of airborne imagery. In the case of a natural protected area,
the managers often already have this information. As an alternative, information from Spatial Data Infrastructures (SDI) can be obtained and used, although
all data cartographies should be in the same projection.
(ii) Field-based cover estimates of plant species are used to validate the plant species maps generated by processing the imagery. The most reliable way to collect plant species cover over heterogeneous ecosystems is by stratified random
sampling (Canfield 1941). Moreover, to ensure better results for future programs that monitor species distributions, the sampling plots must be located in
as many permanent sites as possible. The number of measurement plots is
determined by the number of plant communities, the degree of biodiversity
present, and the heterogeneity of environmental factors impacting the survey
area. For validation purposes, the plot size is determined by the spatial resolution of the imagery, typically being three times the pixel size of the acquired
imagery. To measure plant species cover in the sampling plots, a quantification
method independent of vegetation type is desirable.
(iii) A spectral library is generated by a dedicated field spectroscopy campaign, to
characterise the spectral reflectance response of the plants and to obtain the
endmember for each species. There is no standard protocol for generating a
spectral library for plant species (Pfitzner et al. 2010). Nevertheless, the measurement protocol for the spectral library must combine a sampling strategy
and observation procedure for the spectra acquisition of the canopy. In addition, all the aspects of spectra processing must comprise the spectral reflectance files preparation, spectral library generation and separability quantification
between plant species presented. The sampling strategy is better accomplished
by stratified sampling, and must consider obtaining several acquisitions during
the phenological cycle to assess the optimum time of the year for separating
the species. The aim of separability analysis is to estimate ranges of spectral
variability within species and the spectral similarity between species.
M. Jiménez and R. Díaz-Delgado
Field Data
Field data is needed from the study area describing the plant communities present
as auxiliary information for planning the airborne campaign, supporting the imagery analysis, and validating the cartography generated by the process. The user
organization plays a key role in the collection and preparation of these field data.
(i) The topographic and plant community maps of the study area are used to plan
the hyperspectral flights and provide cover estimates of the plant species. A
Digital Elevation Model (DEM) with high spatial resolution is needed to support the processing of airborne imagery. In the case of a natural protected area,
the managers often already have this information. As an alternative, information from Spatial Data Infrastructures (SDI) can be obtained and used, although
all data cartographies should be in the same projection.
(ii) Field-based cover estimates of plant species are used to validate the plant species maps generated by processing the imagery. The most reliable way to collect plant species cover over heterogeneous ecosystems is by stratified random
sampling (Canfield 1941). Moreover, to ensure better results for future programs that monitor species distributions, the sampling plots must be located in
as many permanent sites as possible. The number of measurement plots is
determined by the number of plant communities, the degree of biodiversity
present, and the heterogeneity of environmental factors impacting the survey
area. For validation purposes, the plot size is determined by the spatial resolution of the imagery, typically being three times the pixel size of the acquired
imagery. To measure plant species cover in the sampling plots, a quantification
method independent of vegetation type is desirable.
(iii) A spectral library is generated by a dedicated field spectroscopy campaign, to
characterise the spectral reflectance response of the plants and to obtain the
endmember for each species. There is no standard protocol for generating a
spectral library for plant species (Pfitzner et al. 2010). Nevertheless, the measurement protocol for the spectral library must combine a sampling strategy
and observation procedure for the spectra acquisition of the canopy. In addition, all the aspects of spectra processing must comprise the spectral reflectance files preparation, spectral library generation and separability quantification
between plant species presented. The sampling strategy is better accomplished
by stratified sampling, and must consider obtaining several acquisitions during
the phenological cycle to assess the optimum time of the year for separating
the species. The aim of separability analysis is to estimate ranges of spectral
variability within species and the spectral similarity between species.
M. Jiménez and R. Díaz-Delgado