DETERMINATION OF VEGETATIVE CANOPY PARAMETERS FROM OPTICAL MEASUREMENTS

ANDRES KUUSK

Institute of Astrophysics and Atmospheric Physics, Estonian Academy of Sciences, Tõravere, Estonia, EE2444

Remote Sens. Environ., 1991, 37(3): 207-218.
Submitted 6 August 1990.

ABSTRACT

Special measurements of barley canopy directional reflectance in the red and NIR spectral regions during vegetative growth were carried out to test the Nilson-Kuusk canopy reflectance model. The model reproduces adequately the measured angular distribution of red reflectance. Systematic differences are observed in the NIR spectral region. Using a simple inversion procedure, the canopy optical and structural parameters are determined from optical measurements in one or two spectral regions. Results of inversion agree reasonably well with those of phytometric measurements.

SUMMARY

The field measurements and data analysis in the present paper were addressed to

(1) investigate experimentally the high resolution directional reflectance angular distribution of a barley canopy, and its changes during vegetative growth,

(2) compare the measured CR distributions and model calculations using the Nilson-Kuusk reflectance model (1989),

(3) estimate barley canopy optical and structural parameters from optical measurements by inversion of CR model.

During vegetative growth, significant changes take place in the structure of the barley canopy. However, little changes occur in the optical parameters. So, the main interest was the esti- mation of canopy structural parameters. The paper demonstrates the moderate success of CR model inversion, and confirms the results from the theoretical analysis of Kuusk and Nilson (1989) that the leaf size and orientation parameters may be best determined in the red spectral region. The Nilson-Kuusk CR model (1989) reproduces adequately the measured angular distribution of red reflectance, whereas in the NIR region there are systematic differences. The data analysis showed that the leaf hair index may be considered constant, at least within one species, as it has done by Vander- bilt and Grant (1985). So, the number of model parameters which should be determined by inversion may be reduced to 8.