A strategy to integrate a priori knowledge for an improved inversion of the LAI from BRDF modelling

  • Authors:

  • Guangjian Yan;Xihan Mu;Yinchi Ma;Zhao-Liang Li

  • Affiliations:

  • State Key Laboratory of Remote Sensing Science, School of Geography, Beijing Normal University, Beijing, China;State Key Laboratory of Remote Sensing Science, School of Geography, Beijing Normal University, Beijing, China;State Key Laboratory of Remote Sensing Science, School of Geography, Beijing Normal University, Beijing, China;State Key Laboratory of Remote Sensing Science, School of Geography, Beijing Normal University, Beijing, China,Institute of Geographical Sciences and Natural Resources Research, Beijing, China

  • Venue:
  • International Journal of Remote Sensing - Recent Advances in Quantitative Remote Sensing: Papers from the Second International Symposium, 25th-29th September 2006, Torrent, Spain
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

We propose a strategy to construct a priori knowledge in Bidirectional Reflectance Distribution Function (BRDF) model-based Leaf Area Index (LAI) inversion. In this strategy, the physical limitations, a best guess and its uncertainty for each parameter needed to be inverted were obtained from a spectral database. Vegetation index (VI) and growth date were used to provide more information about LAI. The relationship between LAI and VI was obtained by forward simulation using the BRDF model. The empirical model of the changing LAI and the growth date was obtained by statistical analysis of more than 600 field samples from a wheat paddock. A SAIL-reflectance model including the hotspot-effect (SAILH model) was used to generate bidirectional reflectance distribution. Gaussian distributed random noises were added on the reflectance as 'observation'. SAILH model was inverted to validate the effectiveness of this strategy. It was further validated using both of the ground measurements and airborne remote-sensing data. It is found that a priori knowledge is important for successful inversion, and our strategy is expected to yield more reasonable spatial and temporal LAI distribution.