The aim of this work is the development and implementation of a new fast and stable algorithm able to create maps of some key vegetation paramenters, which contribute to describe the status of vegetation canopy. The procedure will generate images of retrieved LAI, FaPAR and soil reflectances, in addition to the usual NDVI maps which are immediately calculated by NIR and RED reflectances. Model will analyse data acquired by CHRIS sensor, an experimental hyperspectral high spatial resolution radiometer, on board of PROBA platform. It is reasonably supposed that such analysis will allows us to test the quality and accuracy of images. This is an important aspect for two reasons: the former is that, as previously said, images are acquired by an experimental sensor, which can show istortion in optics, electronic or, simply, in calibration procedure. The latter is that when reflectance data are used to estimate vegetation parameters atmospherical correction algorithms have to be applied in order to have top of canopy reflectances. These algorithms could amplified bias and/or distortions in original data. The procedure will be devoted to retrieve coniferous vegetation parameters. The choice of this particular biome is the almost total absence of a data-base of single needle physical and geometrical properties, which are needed for the vegetation radiative transfer problem solution. Then it is useful to describe needle as better as possible. LOPEX 2003 is one of the most used vegetation spectral characteristics data-base. It provides measures acquired on a large number of canopy types. Broad leaf species are well and accurately described, while coniferous properties are measured on a compressed pastille of a large number of needles. It is obviuos that this is a too strong and unrealistic approximation of needle-leaves spectral characteristics. In the last years, in the framework of BOREAS project, a new radiative transfer model and spectral data-base for coniferous canopy is under development. For these reasons, in this work, laboratory acquisitions will aim at extracting single leaf properties of maritime or domestic pine. As single vegetation element we will consider little pine branches, with a low number of pine shoots. Reflectances and destructive LAI measurements will give us the possibility to estimate "single leaf" properties, like extinction coefficient and reflectance of infinitely thick canopy, then used as input in the retrieval model. This model will be based on the two fluxes approximation. Finally, once LAI, soil reflectances and FaPAR results are validated they can be used in other algorithms to estimate NPP or as input of climate and large scale ecosystem model.
Cristofori, S. (2005). Estimation of coniferous vegetation parameters by hyperspectral high spatial resolution satellite images.
|Titolo:||Estimation of coniferous vegetation parameters by hyperspectral high spatial resolution satellite images|
|Data di pubblicazione:||22-nov-2005|
|Anno Accademico:||Aprile 2005|
|Settore Scientifico Disciplinare:||Settore ING-IND/11 - Fisica Tecnica Ambientale|
|Tipologia:||Tesi di dottorato|
|Citazione:||Cristofori, S. (2005). Estimation of coniferous vegetation parameters by hyperspectral high spatial resolution satellite images.|
|Appare nelle tipologie:||07 - Tesi di dottorato|