Chlorophyll concentration mapping with MIVIS data to assess crown discoloration in the Ticino Park oak forest

Airborne hyperspectral remote observations, characterized by high spatial and spectral resolution, allow the estimation of quantitative vegetation variables useful in forest condition assessment. In this research, total chlorophyll (a + b) concentration (Cab), a biochemical variable describing crown discoloration rate, was mapped to assess oak (Quercus robur L.) condition in the Ticino Regional Park. A simulation experiment was conducted to evaluate the error in Cab estimation due to ecological variables (i.e. canopy leaf area index and understorey characteristics) and to sun-sensor configurations when optical indices are used. Canopy reflectance was simulated by means of the PROSPECT leaf radiative transfer model (Jacquemoud and Baret 1990) coupled with the SAILH canopy radiative transfer model, a variation of the SAIL (Scattering by Arbitrarily Inclined Leaves) model modified to include the hot spot effect (Verhoef 1984, Kuusk 1991). The vegetation was modelled as a two layer medium with oak canopy as the top layer and the understorey as the bottom layer. Simulations were performed for varying leaf Cab and canopy Leaf Area Index (LAI) of the top layer, yl (mean leaf inclination angle) and LAI of the bottom layer (LAIu) and sun-sensor geometry. Optical indices were calculated and used in Cab retrieval. Simulations demonstrated that errors in Cab estimation were negligible when MTCI (MERIS Terrestrial Chlorophyll Index) was used, thus indicating that MTCI was the most reliable index in mapping Cab in this forest environment. Empirical models based on optical indices were developed to map Cab from Multispectral Infrared and Visible Imaging Spectrometer (MIVIS) images acquired by an airborne survey on the ark forested area. A regression analysis between Cab concentration measured in leaves sampled in field and optical indices computed from hyperspectral MIVIS data was conducted. The MTCI index showed the highest performances and was therefore used to map Cab concentra