Mapping spatial continuous distributions of indicator values using imaging spectroscopy

Continuous maps of plant indicator values can provide spatial explicit information about environmental gradients and processes and are an important input source for ecological models. However, as field-based mapping requires extensive sampling, most assessments are at the plot level instead of covering large areas. The potential of remote sensing to map Ellenberg indicator values has been shown in earlier studies, however these were based on a relatively small sampling set in a grassland ecosystem.

In the present study we investigate the spatio-temporal robustness of the relations between imaging spectroscopy derived vegetation reflectance and plant indicator values. Ellenberg indicator values for water supply, soil pH and soil fertility were derived using species data from vegetation plots of a floodplain (n=400) and dune ecosystem (n=150) in the Netherlands. For the two ecosystems, airborne imaging spectrometer data were acquired using the HyMap sensor in 2004 and the AHS sensor in 2005, respectively. Partial least-squares regression was adopted to relate indicator values and plot reflectance extracted from the imaging spectrometer data. Applying the regression to the imagery resulted in spatial continuous maps for the three investigated indicator values (R² = 0.39 – 0.45 for independent test-set).The results show that comparable reflectance wavelength regions are important to estimate indicator values for the two investigated ecosystems. However, intercalibration of the regression models shows that they are partly depending on vegetation and sensor type. Further research will focus on the development of a mechanistic approach to model the relation between vegetation canopy reflectance and patterns of indicator values.