area of 
N. The 
erately 
sult of 
e 1S its 
d et al 
fringe, 
:slands 
of the 
. Water 
5 dairy 
>s plus 
". John 
1984, 
) were 
n was 
led. In 
uction 
n was 
IS Was 
| using 
S 2.1, 
en the 
based 
1 rules 
nption 
hange 
ned to 
Tus to 
From 
n ina 
n was 
sk for 
1e risk 
| et al. 
Bemigisha, Jane 
  
X; - (Ri - X min) / (Xpmax“ X in ) (1) 
Where: X= standardised pixel 
R= raw score (raw pixel value) 
Xmax is the maximum pixel value 
Xmin is the minimum pixel value 
According to the criteria, the nearer the papyrus pixel to a land cover, the higher the potential risk, so, the Xmin was 
translated into the highest risk, while the Xmax was translated into the lowest risk. The index was used to slice 
(reclassify) the land cover type distance to papyrus maps into regular 10 classes showing respective risk levels from 1 
(low level), to 10 (high level). 
2.3.2 Criteria 2: Papyrus change in area (ha) by the other land cover types: A land cover that had higher area 
(ha) overtaken from papyrus was assumed to have a higher potential risk to the swamp. The land cover change maps 
1967/1984, and 1984/1995, the areas calculated for each land cover type according to the change to papyrus ( positive or 
gain for papyrus)or from papyrus (negative or loss to papyrus) were aggregated to get the total change to papyrus area 
between 1967 and 1995. 
2.3.3 Criteria 3: Change rate in area(ha) per year of the individual land cover types: It was assumed that if a 
landcover type was increasing at a faster rate, it had a higher threat tp papyrus area. Area(ha) for each land cover type 
was calculated for 1967, 1984, and 1995 from the respective land cover maps in ILWIS using the HISTOGRAM 
function . The histogram table was imported into excel for calculating the change rate using the following formula: 
(y2-y1)/t (2) 
Where: 
y1 the initial year, 
y2 the final year 
t is the difference between y2 and y1 
2.4 Multicriteria Evaluation and Modelling 
To distinguish between human impacts and more natural influences, two scenarios were developed: 
— Scenario 1: impact of land cover types, including open water as most influential factor 
— Scenario 2: impact of land cover types, excluding open water 
241 Weights of land cover types by the criteria and their ranking : For the criteria 2 and 3, to be used in 
Scenario 1 and 2, a set of. weights were derived using the Regime multiple criteria evaluation module of DEFINITE, a 
Decision Support System (Janssen and Herwijnen, 1994). A decision or effect table was created. For criteria 
comparison purposes a normalisation (standardisation of the scores) is required, and this was done by quantitative ratio 
scale (Eastman et al 1995) before the evaluation process as: 
standard score (i)= raw score(i)/row maximum (3) 
The standard score is the new criterion score standardised, 
The raw score is the criterion score before standardisation 
The row maximum is the maximum criterion score by the same row. 
Weight assignment to each criterion and finally ranking of the land cover types from lowest to highest risk was done 
using the Regime method, which is based on the following formula: 
  
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000. 167