Waste land in 
ential future 
as active 
are included. 
Df how this 
eturns. 
>w differing 
n figures. 
74 1980/82 
93 4779 
29 10547 
51 1428 
11 2445 
/aste land is 
to the WMCC 
y 31% and 45% 
vely. Such 
jfinition and 
revision or 
ration of the 
ew district 
aphs pay heed 
t 'if however 
lue to budget 
.ass if icat ion 
veral stages 
.1 a two-fold 
ireliminary 
graphy being 
i accuracy of 
:ea 
aerial 
il 
and 1980 
serial 
i 
.te imagery 
assess their 
survey maps 
:he WMMC at a 
results of 2 
Ln the WMCC; 
ive (1983) . 
rial Property 
ty Planning 
i survey maps 
n transferred 
ps, together 
Site data is 
3C mainframe 
ut using the 
sets of low 
latic aerial 
:s, flown in 
October 1971 and July 1980 at scales of 
1:12000 and 1:6000 respectively. Both sets 
are of good quality, being flown in late 
morning/midday to minimise shadow effects of 
buildings. However the 1980 set suffers from 
16 missing prints which covered the 
north-west quadrant of the area, therefore 
figures for this area come from the WMCC 
ground survey. 
Cloud free, 30m resolution satellite 
imagery from the Landsat 5 TM is used for the 
second classification stage of this study. 
The image was obtained on 26 April 1984, Path 
205/Row 25. Only 6 channel data was 
requested. Band 6, the thermal channel, was 
not required for this study. The image was 
obtained form the National Remote Sensing 
Centre archive at Farnborough, UK. The 
characteristics of the TM sensor will not be 
discussed here and are well documented in 
IEEE Transactions on Geoscience and Remote 
Sensing, Volume GE-22, Number 3, May 1984 and 
Photogrammetric Engineering and Remote 
Sensing, Volume LI, September 1985, Number 9. 
As yet little analysis of data quality has 
been possible, however from preliminary 
investigations it seems that the combinations 
of channels 1 or 2, with 4 and 5 may produce 
the best results. 
The study area, part of the Dudley 
Metropolitan District Council in the 
south-west of the West Midlands Black Country 
Conurbation was chosen for a variety of 
reasons. The availability of ground survey, 
aerial photography and satellite data 
presented no problem. The area has changed 
dramatically in a short time and features 
extractive and heavy industries, extensive 
rail and canal networks and there have been 
attempts at reclamation. It is relatively 
close to Aston University which facilitates 
field-study and communication with the local 
council. An eight point classification 
modified from the DOE 1974 Derelict Land 
Return System by the WMCC is used for the 
aerial photographic survey. Land without any 
beneficial use is divided into: 
1. Spoil heaps 
2. Excavations and pits 
3. Military and other Service Dereliction 
4. Disused Rail Land 
5. Disused Sewage Works and Installations 
6. Disused Waterways Land 
7. Neglected Waste Land 
8. Other 
The 8 basic characteristics were further 
divided by the WMCC for their ground survey 
to give 23 sub-divisions. As many of these 
are not apparent from aerial imagery they 
will not be exhaustively listed here. 
A photo mosaic was compiled using the 
panchromatic aerial photography. These were 
examined stereoscopically and areas of 
dereliction were plotted on acetate overlays 
placed on alternate photographs within the 
flight strips. The vertical strips overlapped 
by approximately 60% and adjacent photographs 
had a 25% overlap. 
A minimum site mapping unit of .25 
hectares was specified, this corresponds to 
an area of 2.7 TM pixels (TM resolution being 
approximately 0.1 ha). It was considered 
that this would give sufficient accuracy for 
the TM classification, as according to Wilson 
and Thompson (1982), MSS data, with 0.5 ha 
resolution, can be used for map scales down 
to 1:25000. Sites were identified, classified 
according to type and transferred to and 
recorded on 1:10000 clear film maps of the 
area. 
This procedure was followed for both the 
1971 and 1980 photography, leading to a third 
overlay noting change between the 2 dates. 
The resultant polygons were digitised and 
the perimeter length and area of all the 
sites was recorded. 
Vegetation maps are currently being 
produced by the same method, using a 
different system, to aid in the satellite 
classification. 
The aerial classification map for 1980 
will be used in conjunction with the 1980 and 
1984 WMCC ground survey maps of derelict and 
waste land and vacant industrial property. 
These maps will be amalgamated and used to 
choose training areas for the computer 
classifier. This will also provide a means 
of visual comparison with hard copy of the 
satellite classification. 
Computer classification of the TM data is 
the next planned stage in the research. 
Preliminary work has involved demarcation of 
the study area, requiring the definition of a 
167 x 167 pixel extract from a 512 x 512 
sub-scene. It is the intention to classify 
only the derelict and waste sites via a 
thresholding exercise. Training sets will be 
identified from the aerial photography 
reference maps, whilst in order to maximise 
classification accuracy only large and 
relatively homogenous areas will be used. 
Areas of urban open space other than spoiled 
land, such as golf courses and parks, will be 
masked so they do not interfere with 
classification. The reference maps will also 
be used as a measure of classification 
accuracy. 
It is intended to investigate the further 
incorporation of the classification, land use 
and any ancillary data into a Geographic 
Information System (GIS), this together with 
the potential of satellite imagery for urban 
studies is discussed later. 
Analysis of aerial photograph classification 
The results of the aerial photographic survey 
of derelict and waste land are presented in 
Tables 2, 3, and 4. The most meaningful 
results are those in Tables 2, dealing with 
the number and size of derelict sites in the 
mapped area. 
In 1971, 571 out of 2500 ha, 23% of the 
map area were derelict. This figure had been 
reduced in 1980 by 119 ha, or 5%, to 18% of 
the map area, 453 ha. The reduction in size 
of the total derelict area is very 
encouraging, reflecting the WMCC success with 
reclamation. However, a more worrying trend 
is the number of sites that are derelict in 
the Dudley area. The increase in site 
number by 59% from 117 in 1971 to 185 in 1980 
suggests a more disseminate distribution. 
The number of sites in the 0-0.5 ha and 0.5-1 
ha range has increased by 69 between the two 
dates and 48% of all the sites in 1980 fit 
this category. 
The reason behind the rise in site number 
is an acceleration of the decline in 
traditional manufacturing industries, which 
reached epidemic proportions in the early 
1980s. Cave (1984) reveals that the amount of 
vacant industrial floorspace in the Black 
Country increased by over 400% from 60 ha in 
April 1979 to 258 ha in April 1982. 16 ha of 
this is in Dudley. For the same period 
unemployment rose 290% from 77,000 to 
225,000. It seems reasonable in the light of 
this evidence to conclude that the closure of 
factories and other premises accounts for the 
disproportionate rise in small derelict 
sites.