Please use this identifier to cite or link to this item:
|Title:||Vegetation biophysical variable retrieval using object-based inversion of hyperspectral CHRIS/Proba data|
|Authors:||ATZBERGER Clement; RICHTER Katja|
|Citation:||Proceedings of the Hyperspectral 2010 Workshop - ISBN 978-92-9221-247-6, ISSN 1609-042X vol. ESA SP-683, May 2010|
|Publisher:||European Space Agency (ESA)|
|Type:||Articles in periodicals and books|
|Abstract:||The accurate retrieval of biophysical vegetation characteristics is required in a variety of applications including agriculture, forestry, meteorology, climate modelling and ecology. The most stable and accurate approach for the estimation of vegetation variables such as leaf area index (LAI) or chlorophyll content is based on the inversion of radiative transfer models (RTM). This technique has proven to overcome critical limitations of traditional methods, based on empirical relationships between the vegetation characteristics and vegetation indices (VI). However, the use of RTMs is seriously hampered by the well known ill-posed inverse problem, mainly caused by the under-determined nature of the modelling schemes. Different regularization strategies have been exploited in this regard, for instance the use of prior information to change the cost function or to constrain the parameter ranges. In this study, the object-based inversion approach is extended and applied on hyperspectral multi-angular CHRIS/Proba imagery. The method builds on the geo-statistical principle that biophysical characteristics of adjacent pixels are generally more similar than those further a part. The approach was implemented as a two-step inversion based on PROSPECT+SAIL generated look-up-tables (LUT). The LUTs were configured according to the multi-angular properties of CHRIS/Proba. The spectral-directional signatures of neighbouring pixels were simultaneously optimized for that all pixels of a given agricultural field yield a common average leaf angle (ALA). Estimated LAI values were evaluated with a large set of LAI measurements acquired over different crops during the Barrax 2004 campaign (Spain). Results of the object-based retrieval technique were compared with those of a traditional pixel-based LUT. The findings of the study demonstrate that the proposed regularization yields in most cases more accurate and spatially consistent results compared to the pixel-based method. Benefits of hyperspectral and multi-angular information are discussed in this context together with possible improvements and future extensions.|
|JRC Directorate:||Space, Security and Migration|
Files in This Item:
There are no files associated with this item.
Items in repository are protected by copyright, with all rights reserved, unless otherwise indicated.