NR is responsible for two parts of the project, which are described below.
Multi-frequency SAR: The aim of this work is to develop a model for defoliation and leaf area index (LAI) based on SAR imagery. The data available for model development will be TerraSAR X-band data, ENVISAT C-band data and full polarimetric Radarsat-2 C-band data. Ground truth is available from field measurements with an LAI-2000 instrument, estimates of LAI based on LIDAR, as well as detailed data from a number of sample trees. The first phase of the work will focus on an exploratory data analysis investigating the LAI information contents at the different radar frequencies by analysing homogeneous areas of the forest with varying field-measured LAI. The different wavelengths of the SAR should open for measurements at varying penetration depths in the forest. In the second phase of the work a model will be built based on the best combination of SAR frequencies and/or features measured in the SAR data. The robustness of the model will be studied, and recommendations will be made for practical applications.
Hyperspectral data from airborne sensors: The aim of this work is to develop a model for defoliation and LAI based on imaging spectrometer data. The data available for model development will be airborne hyperspectral imagery from NEO HySpex VNIR-1600 sensor with 160 bands in the range of 0.4-1.0 μm. The first phase of the work will focus on an exploratory data analysis for LAI information in the hyperspectral data. The hyperspectral data should allow quantification of the attenuation effect due to the insect defoliation relative to spectral signatures of a healthy forest. Ground truth is available from LAI-2000, LIDAR and detailed data for a number of sample trees. In the second phase of the work, a model will be built for retrieval of LAI applying the results from the first phase. The robustness of the model will be studied and recommendations will be made for practical applications.