133
112
17.6
214
314
21.9
254
28.7
23.9
17.6
12.1
18.3
26.1
y of
50.0
ES
ility of
ig GIS
iccord-
X gen-
n pro-
or this
impor-
) 123 38
quisitions
S-1
, 21-
3etter
din
and to
d XS2
s. The
easing
reases
30-40
ul for
al and
environmental influences affect only, and almost in the same
amount, the absolute mean level of radar reflectivity of forest age
classes.
Capabilities of SPOT/XS3 and ERS-1 data for discriminating
stands of different ages were examined by calculating the linear
correlation coefficients between forest backscatter and age for
four cases (Table 4):
* correlation between age (all age classes, from 1 to 160 years)
and mean backscatter of forest stands, not differentiating species
composition,
* correlation between age (all age classes, from 1 to 160 years)
and filtered pixel values on forest stands, not differentiating
species composition,
* correlation between age (only age classes from 1 to 30 years)
and filtered pixel values on forest stands, not differentiating
species composition,
* correlation between age (only age classes from 1 to 30 years)
and filtered pixel values on the same stand type (i.e. same species
or species composition).
As expected from Fig. 5, the low correlation coefficients ob-
tained (Table 4) in the first two cases prove that neither SPOT
nor ERS-1 allow complete discrimination of age classes from 1
to 100 years. In the last two cases, if only the first three age
classes (from 1 to 30 years) are considered, then a very high cor-
relation is observed between age and ERS-1 (all acquisitions) or
SPOT/XS3 signals, for either not differentiating species compo-
sition or the example of a special stand type such as red oak.
| PM Pixel
| Based
All Stand Types
1-100 1-100
Years Years
- 0.09 - 0.43
0.15 - 0.20
0.20 0.22
0.25 0.28
0.19 0.16
0.21 0.09
0.16 0.04
0.16 - 0.07
0.18 - 0.03
0.17 0.52
0.15 0.08
0.13 0.19
0.23 0.28
0.12 -0.11
0.18 0.20
Table 4: Correlation Coefficients Between Backscatter of Forest
Age Classes (1-100 Years) and Age in one SPOT and 14 ERS-1
Scenes on per Polygon Base, on per Pixel Base, on per Pixel
Base for Only the First Three Age Classes (1-30) and on per
Pixel Base for the First Three Age Classes of a Red Oak Stand.
The opposite signs of the correlation coefficients between forest
backscatter and age class exhibited by SPOT/XS3 and ERS-1
indicate a possible synergy of both sensors for retrieval of forest
stand age. Thus, a fusion of the data has been tried by simply
USIng a ratio of the SPOT/XS3 image radiometry and the res-
335
caled filtered ERS-1 amplitude. These ratios of SPOT/XS3 to
every ERS-1 acquisition were statistically evaluated by comput-
ing the linear correlation coefficients to age classes. Results are
given in Table 5 for non-differentiating species composition and
for a special stand type (ash). By showing fairly good correla-
tions for forest age ranging from 1 to 100 years, the results in
Table 5 indicate a superior sensitivity of this SPOT/X3 to ERS-1
ratio index to age classes compared to each single sensor.
For All Stand For Ash
Types
1 - 100 Years | 1 - 100 Years
- 0.75 - 0.67
- 0.77 - 0.75
- 0.76 - 0.75
- 0.75 - 0.77
- 0.80 - 0.77
- 0.78 - 0.78
- 0.75 - 0.66
- 0.76 - 0.74
- 0.82 - 0.86
- 0.79 - 0.87
- 0.78 - 0.75
- 0.87 - 0.82
-0.77 - 0.83
- 0.79 - 0.89
Table 5: Correlation coefficients between ratio of SPOT/XS3
and ERS-1 for 14 ERS-1 acquisitions of forest age classes 1-12
and age without differentiation of stand type and of forest age
classes 1-10 and age of an ash stand on per pixel base.
7. SUMMARY AND CONCLUSIONS
The results of this study add to the growing evidence of the valu-
able capabilities imaging radars have for the studying and moni-
toring of forests due to the sensitivity of radar to a variety of
processes in vegetation ecosystems and the proven synergy to
optical satellite data.
The changes in backscatter observed by the ERS-1 SAR have
been confirmed to be related to the variation of meteorological
parameters (i.e. precipitation, temperature) for different landuse.
It confirms that the temporal sequence of the C-band reflectivity
of vegetated areas measured by ERS-1 appear to be sensitive to
changes in moisture conditions as already stated by Hsu et al.
(1993), and effects of temperature on dielectric properties
(Wegmüller et al., 1994). Long term variations due to phenologi-
cal and seasonal effects can then be used to discriminate land use
classes. Temporal backscatter profiles show that a discrimination
of vegetated areas (grassland, agriculture, forest) is achieved
from October to May. In conclusion, high spatial resolution
monitoring of vegetated surfaces, forest conversion into culti-
vated or bare soils, or agriculture management is possible using
multitemporal ERS-1 data.
The results from this investigation demonstrate that the C-VV
band of ERS-1 is sensitive to age (and therefore closely related
parameters such as tree height, woody biomass) in young decidu-
ous and coniferous forest stands of temperate zones. This capa-
bility can be important for forest ecology studies. First, informa-
tion about the population and growth dynamics of successional
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996
