816 
this stage and require further research in the 
application of various classification routines, it 
has shown that the introduction of radar backscatter 
either with Landsat data or from different incident 
angles does tend to separate classes that were 
previously difficult using Landsat alone. Concern 
at the range of backscattered response values from 
individual urban surfaces has been brought out, with 
the suggestion that mean values over a cell with lower 
spatial resolution may be more appropriate. This 
would also overcome the relatively high registration 
errors between the radar scenes, which were of the 
order of one and two pixels along and across track. 
Smoothing after registration would reduce these to 
0.3 and 0.7 of a pixel if a three by three mean filter 
were used, which would be more comparable with the 
errors of 0.5 pixel obtained when different Landsat 
scenes are registered. 
Not considered in this paper, but an area of further 
research, is the use of the variability of the data as 
a positive factor in classification. As can be seen 
from the scattergrams the synthetic urban surfaces 
of road and building, when compared to the natural 
surfaces of forest and soils, have a greater vari 
ability. This textural information, as measured by 
standard deviation or derived using a high pass 
filter, may contain additional means of separating 
urban classes. Further work also needs to be under 
taken on the relationship between building spacing 
and backscattered response, and the effect of change 
in incident angle on the radar view, as projected 
at right angles to the incident radiation, of built 
urban surfaces. 
REFERENCES 
Bryan, M.L. 1979. The effect of radar azimuth angle 
on cultural data. Photogrammetric Engineering and 
Remote Sensing. 45:1097-1107. 
Hardaway, G. & G.C. Gustafson 1982. Cardinal effect 
on Seasat images of urban areas. Photogrammetric 
Engineering and Remote Sensing. 48:399-404. 
Symposi 
An an; 
E.C.Hyat 
Remote Sen 
Introduct 
The aims 
of aeri< 
studies 
County ai 
the poter 
problem. 
The e' 
County B 
District 
of the 
change d< 
derelict 
dense urt 
A remc 
monitorii 
(TM) ima< 
critical 
findings 
satellite 
Derelict! 
Birminghc 
area in 
percenta 
Departme 
Survey. 
(1975) c 
the Wes 
classifi 
in a la 
survey : 
nearly 5 
ha, co 
Represen 
or 26%, 
increase 
District 
Wolverhai 
Accor 
land bee 
that be! 
(1977) £ 
derelict 
ha/year. 
the ca 
consider 
centre 
networks 
The 
extracti