The objective to derive accurate estimations of timber
volume and forest biomass using SAR data (Guindon et ai,
1981) ean only be reached without implementation of a DTM
and radargrammetric corrections respectively, if the terrain
is flat. For mountainous regions, which - at least for
large parts of the world - show most forests, radargrammetry
becomes an absolutely necessary tool. Visible and
near-infrared scanner data may assist in getting even more
precise information on plant heights and biomass (Rosenthal
et al. 1985). Here, geometric referencing is of substan
tial importance.
Recent activities in applying multifrequency, multi-
polarization and multi-incidence angle SAR data for land
cover classification, especially for natural vegetation and
crops, yielded promising results (cf. Mehta 1984). Changes
in the SAR viewing geometry give rise to a spatial variation
in land cover mapping accuracy. A high portion of back-
scatter variance can be explained by changes in the
incidence angles (Frost et al. 1984, Foody 1985).
5. CONCLUSIONS
The usefulness of radargrammetry in supporting the
analysis of SAR imagery has been demonstrated in a number of
examples. Radargrammetric techniques, originally developed
to solve the geometric aspects of radar mapping problens,
provide the basic tools for map-guided geodata acquisition,
for change detection leading to topographic and thematic map
revision, and for the integration of multi-sensor remote
sensing data into geographic information systems, and
consequently can play an important role in geosciences.
Whatever application, all studies utilizing digital SAR
data still demand some knowledge about radargrammetric
details. To attract a larger community of experimenters,
the information derivable from SAR imagery should be
presented in an easily exploitable form, i.e. data pre- and
postprocessing should better meet the ultimate users' needs.
This, in turn, requires major efforts to improve the
understanding of potential SAR data applications.
6. ACKNOWLEDGEMENTS
The present paper results from joint activities of the
Graz Research Center radar group. The authors thanks in
particular go to D. Strobl and J. Raggam.
330
-— PA PS an C» rn cC. mee C) dT Oo TI 3 D Ul DH 3)
ae GF OX um d