Forest canopy chemistry with high spectral resolution remote sensing

• Published in: International journal of remote sensing 1996-04, Vol.17 (6), p.1107-1128
• Main Authors: ZAGOLSKI, F., PINEL, V., ROMIER, J., ALCAYDE, D., FONTANARI, J., GASTELLU-ETCHEGORRY, J. P., GIORDANO, G., MARTY, G., MOUGIN, E, JOFFRE, R.
• Format: Article
• Language: English
• Subjects: Animal and plant ecology; Animal, plant and microbial ecology; Biological and medical sciences; Forestry; Fundamental and applied biological sciences. Psychology; General forest ecology; Generalities. Production, biomass. Quality of wood and forest products. General forest ecology; Sciences of the Universe; Synecology; Terrestrial ecosystems
• ISSN: 0143-1161; 1366-5901
• DOI: 10.1080/01431169608949073

Forest ecosystem modelling requires information about canopy chemistry. This is usually obtained through chemical analysis and laboratory spectrometric measurements. The potential of spectrometric remote sensing was investigated with two airborne campaigns organized in 1991 with AVIRIS (Airborne Visible/Infrared Imaging spectrometer) and in 1993 with ISM (Infrared SpectroMeter) over the 'Landes' forest (south-west France): AVIRIS covers the 400-2500 nm spectral range with 210 bands, whereas the ISM instrument is an airborne profiling spectrometer that operates in the 800-3200 nm spectral range with 128 bands. The study area consists of homogeneous parcels of maritime pines with a wide variety of ages from 2 to 48 years. Simultaneously with the airborne acquisition, foliar samples were collected in the field. These samples were chemically analysed for determining nitrogen, lignin and cellulose contents. Reflectance spectra of dried pine needles were obtained with the help of two laboratory spectrometers: (1) the Technicon lnfraAlyser-450 with 19 spectral bands centred on chemical absorption features; and (2) the NIR-6500 System with l0nm wide 1050 bands from 400 nm to 2500 nm. Predictive relationships of nitrogen, lignin and cellulose concentrations were established by using stepwise regression analysis on the laboratory spectral measurements. These predictive relationships were quite different, depending on the laboratory spectrometers and the year of sampling. Consequently, different correlations (r 2 ) were obtained between predicted and actual chemical concentrations: 66-94 per cent for nitrogen, 37-79 per cent for lignin and 45-85 per cent for cellulose. The stability of predictive relationships from laboratory to remote sensing level was especially analysed. The application of laboratory derived predictive equations to airborne data led to encouraging results: best correlations (r 2 ) were obtained for nitrogen (AVIRIS: 55 per cent -ISM: 66 per cent) and cellulose (AVIRIS: 63 per cent) but lignin could not be predicted. It was attempted to improve these results while talking into account atmospheric effects: whereas AVIRIS-derived correlations were not improved, ISM-derived correlations were improved for nitrogen from 66 per cent to 76 per cent and lignin from 9 per cent to 77 per cent. The better signal-to-noise ratio of ISM may be the reasons for the better results obtained with this instrument