Seasonal spectral variation of Zostera noltii and its influence on pigment-based Vegetation Indices
Résumé
The influence of phenological variations on the reflectance of the intertidal marine angiosperm Zostera noltii was studied using spectroradiometry and pigment analysis. Leaves were sampled each month from March to November and spectral reflectance was measured in the 400-900 nm wavelength range in the laboratory for increasing biomass. High Performance Liquid Chromatography (HPLC) was used to assess the pigment composition and concentration of the leaves and their epiphytes. Several vegetation indices (VIs) were tested for their sensitivity to seasonal variations, from NDVI-like ratios to indices based on derivative analysis. Pigment concentration showed a seasonal effect, with the highest concentrations observed in summer, synchronous with the above-ground biomass peak. The seasonal variations were clearly visible in the leaf reflectance spectrum, showing for the same amount of vegetation, lower reflectance in summer for the visible wavelengths and a higher near-infrared (NIR) plateau compared to the other months. Indices based on the difference between the red and NIR reflectance values, typically all the NDVI-like ratios, were the most sensitive to seasonal variations in pigment concentration. In fact, seasonal effects were systematically reduced for VIs including a blue band correction, namely the modified Specific Ratio (mSR(705)), the modified Normalized Difference (mND(705)), the modified NDVIs for Landsat, FORMOSAT and narrow band NDVI (mNDVI(673)) as well as the Atmospherically Resistant Vegetation Index (ARVI). However, these indices showed a faster saturation for increasing seagrass biomass, except for the mSR(705), which was the least sensitive to biomass saturation. The interesting properties of this index suggest that it should now be tested for airborne or satellite remote sensing mapping of Z. noltii beds but it requires a high spectral resolution and cannot be applied to multispectral satellite images.