3.2 Ware 
4. CONCLUSIONS 
The conditions at the Ware site were different in that 
landfill gas was migrating laterally and affecting both 
forested and agricultural areas adjacent to the site. Thus 
there was not the same degree of inherent variation in 
soil conditions present at the Panshanger site which 
could induce variations in plant growth and mask the 
effects of landfill gas. 
Damage classification of the forested area was carried out 
on a crown-by-crown basis using the colour infrared 
aerial photographs and a classification system (Table 2) 
developed from several sources (Murtha, 1976; Ciesla et 
al , 1985; Groves, 1989). An obvious pattern of 
damage was present with those trees situated between the 
site and agricultural fields to the west of the site 
showing damage symptoms most frequently. High 
levels of landfill gas had been recorded in this wooded 
area. 
Supervised classification of the woodland was also 
attempted, using the results of the manual classification 
to identify training areas. To date, this approach has not 
shown much success. 
The agricultural fields to the west of the site were 
examined using the band combinations given above. 
Dead and unhealthy vegetation was observed to exist in a 
pattern generally parallel to the edge of the site and 
known to have high concentrations of landfill gas in the 
subsurface. 
3.3 Future Work 
The next step in the work is the supervised calssification 
of both Daedalus ATM and video imagery, followed by 
correlation of aerial imagery and gas concentration data. 
The usefulness of the aerial video data for this type of 
application will be assessed and the performance of the 
two systems compared. 
The results to date suggest that, under certain conditions, 
remote sensing may provide a useful and timely means 
of detecting and monitoring the environmental impact of 
landfill gas. However, due to the fact that the effects of 
gas on vegetation growth may be masked by a number of 
other environmental variables, each situation must be 
assessed individually. The likelihood of remote sensing 
techniques being successful depends on sufficient 
spectral variation existing between healthy and unhealthy 
vegetation, and on the ability to distinguish between 
stress caused by landfill gas and that induced by other 
factors. In summary: 
1. Sufficient quantities of gas must exist in the root zone 
such that the plants are stressed, and that this stress is 
manifested as a change in the spectral properties of the 
plant. 
2. Relatively homogeneous ground cover must exist. 
Variations in soil type or qualitiy, agricultural practices, 
etc., may result in variations in vegetation health that are 
unrelated to landfill gas. 
Initial comparison of ATM and video data indicate that 
the Daedalus scanner provides additional spectral 
information, particularly in the middle infrared. 
Although the initial costs of data collection are high 
compared with the video system, this may be partially 
offset against the time involved in preprocessing of the 
video data. The cost of video data collection is mush 
reduced; additional advantages are the lower cost of data 
processing, improved spatial resolution, and increased 
convenience of data collection. 
Manual interpretation of large scale colour infrared 
photographs will continue to find use in this type of 
study, although due to the time and skill involved in 
undertaking this type of analysis, its use may be 
somewhat restricted. 
In conclusion, in the light of recent legislation in Great 
Britain there is an urgent need for a cost-effective and 
timely method of collecting information about the lateral 
extent of migration of landfill gas. The initial results 
indicate that remote sensing has a role to play; the 
important conclusion will be which system can fulfill the 
required role in the most cost-effective manner. 
Classification Rating 
Textural Properties of Crown 
Spectral Properties of Crown 
0 (Healthy) 
Irregular perimeter, inner crown 
not visible, branches not visible 
Even colour, deep red 
1 (Sickly) 
Sparse crown periphary 
May be some mottling at 
edges, colour pink-red 
2 (Sick) 
Thinning of crown, inner crown 
and ends of branches visible 
Uneven colouring, lighter 
pink hue, pale ’flags' of 
acutely chloritic foliage 
3 (Very sick) 
Sparse crown, inner crown clearly 
visible, rough crown form 
Light pink, may be increased 
numbers of flags 
4 (Dead) 
Skeleton 
Grey - white 
TABLE 2. Tree Crown Damage Classification System for Hardwoods on 
Colour Infrared Aerial Photographs 
666