ire 
ite 
Cares 
Dor an 
1tly 
ad 
AS , 
at 
ing 
2aling 
ional 
re, 
ing 
uld 
ended 
the 
h low 
ted 
y out 
ey 
    
  
The test area chosen for this study was the northern half of the Parish of 
St Catherine (Figure 1) situated to the north-west of the capital Kingston. 
This study area had fairly uniform rainfall and contained all the major soil, 
land use and physiographic types existing on the island. 
The 1:25 000 scale aerial photography was interpreted using a mirror stereo- 
scope and the classification in Table 1, the derivation of which is discussed 
in Collier and Collins (1980). An example of the photographic key developed 
is given in Figure 2. After discrete interpretation the data was transferred 
using a Bausch and Lombz 'Zoomtransferscope' onto an overlay of the existing 
1:12 500 Topographic maps. Obviously a simpler transfer technique such as 
graphical transfer, could equally well have been used. The land use overlays 
were then composited with data on soil type and slope classes derived from the 
original soil survey field sheets (Vernon and Jones, 1958) to provide a map 
having discrete parcels of a single soil type, slope class and land use. A 
change in any one of these three variables necessitated the creation of a new 
discrete parcel. The area covered by each parcel was then measured and the 
areas of the parcels with the same characteristics were aggregated. Measure- 
ment and aggregation were carried out using a digitiser linked to a mini 
computer, but more labour intensive techniques, such as planimeter or dot 
counting, would be equally suitable and probably more relevant to the situation 
in Third World conditions. 
The aerial photographs were then interpreted for eroded land. The eroded areas 
were mapped on the existing land use overlays and measured, recording the soil 
type, slope class, and land use of each area. Eroded areas having the same 
soil type, slope class and land use were also aggregated. 
It was then possible to process all the data to produce lists (Tables 2 and 3) 
of the most heavily eroded combinations of soil, slope and land use. 
In addition to looking at the worst cases, it was also necessary to look at all 
the data in order to study the overall effects of erosion on the different 
soils, slopes and land uses. In trying to analyse the data, however, a number 
of problems occurred. The first of these was due to the soil type and land use 
codes being non-parametric. This rules out the use of most of the standard 
tests for significance. 
It was therefore decided, in the first instance, to carry out cross tabulations 
which would at least indicate whether any patterns were apparent within the 
data. The first result to emerge from this procedure was that there was little 
evidence of a pattern as no relationship was evident between slope class and 
the amount of eroded land. Certainly the work of Hudson (1977) had led to the 
belief that slope would play only a minor role in determining the occurrence of 
erosion, but even so the lack of correlation between slope and erosion was 
still surprising. The indications were that erosion was a greater problem on 
the flatter lands! Almost certainly this was a function of them having the 
better soils under the most intensive cultivation. 
Although some patterns were discernable in the cross tabulations of the soils 
with erosion, and land use with erosion, it soon became evident that a close 
correlation between soil type and land use would make interpretation of the 
results rather difficult. It was found that certain soils were heavily eroded, 
but that these soils were only used for certain crops which were in turn, 
characterised by heavy erosion - irrespective of soil type. 
Table 4 shows that for the soils represented in Tables 2 and 3 between 73 and 
100$ of all erosion is accounted for by the land uses listed in Tables 2 and 3. 
9 39 
—Ó