KEY 
R - Residential areas 
I - Industrial 
S- Service 
W Water bodies 
— Canals 
Fig 1. Map of the land-use in the Blackbrook 
Valley 
4 DATA EXTRACTION - THE CLASSIFICATION 
The photographic interpretation was based on 
a habitat classification devised for this 
project. Habitat mapping is well suited to 
remote sensing as an habitat unit has a 
larger surface area than individual species 
communities and therefore is more easily 
identified. The classification is divided 
into four groups: 
1 land with no cover, 
2 land with man-made cover, 
3 land with water cover and 
4 land with vegetation cover. 
In this paper only land with vegetation cover 
will be discussed. The vegetation categories 
are based on the structure and density of the 
dominant growth forms. The distinction 
between open and closed woodland and 
shrubland categories were dictated by the 
distance between foliage canopies (see 
Fosberg and Peterken 1967). 
28 Tree cover +3m height 
29 Broadleaf communities 
30 Single tree 
31 Closed woodland 
32 Open woodland 
33 Linear woodland 
34 Woody shrub 
35 Coniferous communities 
36 Single tree 
37 Closed woodland 
38 Open woodand 
39 Linear woodland 
40 Woody shrub 
41 Shrub cover -3m 
42 Broadleaf communities 
43 Closed shrub 
44 Open shrub 
45 Linear shrub 
46 Coniferous, spikey communities 
47 Closed shrub 
48 Open shrub 
49 Linear shrub 
50 Herbaceous cover 
51 Ruderal communities 
52 Tall herb and fern communities 
53 Rough tall grassland communities 
54 Wetland herbaceous communities 
55 Tall fragmentary marginal communities 
56 Smooth turf grassland (unmanaged) 
57 Smooth turf grassland (managed) 
58 Rough turf grassland 
59 Floating vegetation 
60 Submerged vegetation. 
Figure 2. Classification categories 
Although the idea of an universal legend is 
attractive no standard classification exists 
for ecological or urban surveys therefore 
new legends need to be devised for individual 
projects to ensure the detail necessary is 
obtained. The vegetation communities 
identified by the N.C.C. were the basis of 
this habitat classification, allowances being 
made for the difference in the survey's 
resolution. 
The accuracy of the photographic 
interpretation and mapping was checked for 
1981 and 1984. A pilot test had identified 
the categories which were frequently 
misc1assified for example bracken. 
Previously this had been an individual legend 
category, but due to interpretation 
difficulties it was included in the tall herb 
and fern category. The accuracy level 
achieved for the 1984 analysis was better 
than the 1981 figure. This was a result of 
the time period between the dates of the 1981 
photographs and the field work, 1983. A good 
deal of development had taken place in the 
area in this period. The 1981 interpretation 
was 89.5% accurate and the 1984 
interpretation was 97.6% correct. The 
statistical tests (see Ginevan 1979 and 
Arnold 1985) suggest the number of errors 
did not occur by chance. 
5 DATA ANALYSIS 
In the literature, studies using air 
photographs to predict bird population 
numbers have not been very successful. 
Various researchers have concentrated on 
predicting bird species numbers, bird 
abundance and bird species diversity using 
variables measured in the field such as area 
of woodland, foliage height diversity, 
foliage canopy density and woodland 
isolation. It was decided to investigate if 
information about the variables used in these 
techniqi 
photogra 
bird cai 
conducte 
Har r i soi 
ornithol 
which we 
approxim 
This fie 
this ana 
One 
variable 
woodlanc 
suggeste 
and Moor 
studies 
linear 3 
bird sf 
woodland 
CO 
LU 
O 
LU 
Cl 
tO 
LO 
O 
Hypot 
co 
W 
o 
LU 
0. 
CO 
Li- 
CD 
O 
z 
ARI 
Spéci 
scali 
from 
FIG 3 
The 
was the 
The area 
Hewlett 
micro coi 
The f: 
bird spe 
breeding 
nesting 
clas sif 
Fifty-s( 
woodlanc 
breeding 
These bi 
classifi