Tal Svoray
Coefficient of determination of 0.9 has been achieved 0.60 7-
for points representing eight relatively homogenous
plots of herbaceous vegetation. When considering the
complex conditions of the study area, this level of
accuracy is satisfactory. However it is important to predicted .
note that in two cases the predicted biomass values |
have suffered considerably high overestimation. The 0:20 |
two points were examined carefully but we could not e
find any particular environmental condition, which we...
may explain their abnormality. This observation will 0.0 010 020 030 040 050 060
be more closely examined with a larger database. If
: : Measured
the analysis of a larger database will show the same
trend, the overestimation could be normalized with an Figure 6. The relationship between predicted and
empirical factor. The case of heterogeneous plots was measured herbaceous vegetation green biomass
much more complex, involved more factors and thus [Kg m™] in homogenous plots.
yields less strong relationship (coefficient of
determination of 0.79). The case of heterogeneous
plots differs from the case of homogenous plots in two
major aspects: first, the range of herbaceous
vegetation green biomass in the case of the
heterogeneous plots is much more limited obviously
as other vegetation formations exist in plots; and
secondly, there is lack of consistency in
overestimation in the case of heterogeneous plots:
over and under estimations seem to be more or less
equal. To summarize this part we may say that it is
Predicted
well believed that albeit the reduction in the 0.00 0.05 0.10 0.15 0.20 0.25
significance of the relationship between the
herbaceous green biomass and ERS-2 SAR
backscatter in the case of heterogeneous plots, it is
still a promising result which may be well improved Figure 7. The relationship between predicted and
with the determination of more precise coefficients measured herbaceous vegetation green biomass
with increasing the database and more attemptions to [Kg m™] in heterogeneous plots.
determine best initial points.
Measured
5 SUMMARY AND CONCLUSIONS
Multi-temporal characteristics of the ERS-2 SAR backscatter from four Mediterranean vegetation formations were
studied in the semi-arid zone of Israel. Significant differences were found between the multi-temporal backscatter
signatures of herbaceous vegetation and woody plant formations. The application of semi-empirical backscatter model
in conjunction with cover fractions derived from Landsat TM data enable to adapt successfully the water-cloud model
to the Mediterranean environment. Significant correlation was observed between the predicted and estimated values of
herbaceous green biomass in homogenous and heterogeneous (mixed) plots. These results imply on the potential of
the ERS-2 SAR data for the estimation of herbaceous vegetation green biomass in wide regions.
ACKNOWLEDGMENTS
We wish to thank the Forest Authority (Keren Kaiemet Leisrael) and the Ministry of Agriculture of Israel for
financing this project.
REFERENCES
Attema, E. P. W. & Ulaby, F. T., 1978, Vegetation modeled as a water cloud. Radio Science Vol. 13, pp. 357-364.
Laur, H., Bally, P., Meadows, J., Sanchez, B., and Schaetter, E., 1997, ERS SAR Calibration — Derivation of the
Backscattering Coefficient Sigma-Nought. In ESA ERS SAR PRI products. Document No: ES-TN-RS-PM-HL09,
ESA web site http://uranus.esrin.esa.it/eeo/lft/ESC2.
Prevot, L., Dechambre, M., Taconet, O., Vidal-Madjar, D., Normand, M., and Galle, S., 1993, Estimating the
characteristics of vegetation canopies with airborne radar measurements. International Journal of Remote Sensing.
Vol. 14, pp. 2803-2818.
326 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part Bl. Amsterdam 2000.