Bemigisha, Jane 
  
LANDCOVER MAPPING AND MULTICRITERIA MODELING FOR EXPLAINING THE SPATIAL AND 
TEMPORAL DISTRIBUTION OF A PAPYRUS SWAMP AT LAKE NAIVASHA, KENYA 
Jane BEMIGISHA 
Uganda Wildlife Authority 
P.O.Box 3530, Kampala, Uganda 
uwapu@imul.com 
KEY WORDS: Landcover, Spatio-Temporal, Change detection, Modelling, GIS, Papyrus Swamp 
ABSTRACT 
This paper is an extract of a research (thesis) done on the distribution of the papyrus swamp at Lake Naivasha in 1998 by 
the same author. Based on area per habitat and trends as indicators, the distribution was derived from the temporal 
analysis of land cover maps. Remotely sensed data (aerial photographs 1967 and 1984 plus a TM satellite image of 
1995) were analysed for the period 1967 to 1995 within two year sets as 1967/1984 and 1984/1995. The distribution of 
the swamp would be explained by the influence of its neighbour landcover. Risk potential zones represented in two 
model scenarios were established indicating relative influence of the landcover based on change rates and distance as 
criteria. The criteria evaluation was done in a Decision Support System and results combined spatially in a GIS. The 
papyrus area declined at a rate of 71 % between 1967 and 1995. The major influence and risk to the swamp area in order 
of magnitude were open water, shrubland, agricultural area, woodland, grassland and built up area. The integration of 
climatic regime, ground water level, soil drainage/depth, topography and anthropogenic factors as grazing and burning 
should be considered for improved explanations. 
1 INTRODUCTION 
1.1 Background 
The distribution of the papyrus swamp at Lake Naivasha has generated interest due to the need for management 
intervention to the problem of the fluctuating swamp area. A decreasing trend spells danger to the surrounding 
communities that derive livelihoods from the swamp. Though there had been observed dynamics involving fluctuations 
and succession, (e.g. Gaudet 1977; Denny 1985; Herper et al 1995), the spatial context had not been evaluated, projected 
and explained for management priority implications. 
The interest in landcover maps for analysis is the wide range of possibilities of their use. This is because of the range of 
surface features that the maps can depict for example water, different kinds of vegetation, rocks, sand and evident man 
made structures. From observed patterns, it is believed that some phenomenal interactions can be deduced ( Leps,1989). 
As such landcover maps have been convenient in ecological studies, for example the AFRICOVER project (FAO, 
1997); identification of hot spots ( Chandria and Surendra 1995) ; and wetlands monitoring ( Yashioka et al 1993). 
Effective management response can be supported by explaining the past and the present and using both for predicting 
the future. This is because within planning and decision making contexts, an explanation beyond the present is desirable 
especially by simplified representations through modelling. In situations where various factors and /or constraints 
interact, the development of criteria for identifying the extent of each one's role in the process can be put into evident 
scenarios to support prioritised management response. The manipulation of multiple criteria has evolved through 
manual mathematical calculations to sophisticated computerised models. These models can be built within Decisison 
Support Systems. The 1990s indicated more of spatial analysis (in GISs) as independent components of the evaluation ( 
e.g. Meijerink et al 1993; Patrono and Oriolo 1995, cited by Patrono 1995), or embedded within the same system as in a 
Spatial Decision Support System (Eastman et al 1995). 
In this study, the objectives were as follows: 
1- To establish the spatial and temporal distribution of the papyrus swamp 
2- To model the potential risk of the landcover types to the swamp area 
3- To assess other factors influencing the distribution of the papyrus swamp. 
With reference to Grillas, (1996) two indicators, surface area per habitat and trends were selected, with the former 
translated into change rates of the surface of the swamp relative to other landcover types while the latter showed the rate 
of papyrus area loss or gain over the time of analysis ( 1967 -1995). The results were used as model inputs to 
produce risk potential zones of the papyrus area through criteria evaluation. 
  
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000. 165