Yuri
Knyazikhin, Boston University, jknjazi@bu.edu
(Presenting)
Dong
Huang, Boston University, dh@bu.edu
Mitchell
Schull, Boston University, mschull@bu.edu
Ranga
B.
Myneni, Boston University, rmyneni@bu.edu
A very strong correlation between multiangle spectral data and canopy height has recently been reported (Kimes et al, 2006). Multivariate linear regression and neural network models developed in this study achieved high accuracies in estimating the canopy height over study sites. Although this result demonstrates the ability of MISR data to predict the vertical structure of forest canopies, the physics behind the observed correlation remains unclear.
The probability that a scattered photon will escape the vegetation canopy in a given direction, the directional escape probability, is a canopy structural parameter that can be derived from milti-angle spectral data. Our analysis of AirMISR and airborne LVIS data suggests that the escape probability and multi-angle spectral data convey the same amount of information about the canopy height. This finding indicates that the canopy spectral invariants explain physics behind the correlation documented in (Kimes et al., 2006). This result will be discussed in the poster.
The MISR operational LAI/FPAR algorithm is parameterized in terms of structural variables that appear in the canopy spectral invariant relationships, i.e., the portion of ground shaded area, the recollision and escape probabilities. As part of LAI/FPAR retrieval, the algorithm estimates these parameters as well as background reflectance, ground cover, LAI and FPAR. Only LAI and FPAR are being archived. We will discuss the feasibility of reliable retrieving the full set of canopy parameters generated by the operational MISR LAI/FPAR algorithm. These parameters can further be used to obtain new information on the 3D canopy structure for use in the CLM, e.g., sunlit and shaded leaf area indices, and ecological models, e.g., the aspect ratio (the ratio of tree height to tree width) and ground cover.