Rob Dekker 
  
MONITORING THE URBANISATION OF DAR ES SALAAM USING ERS SAR DATA 
Rob DEKKER 
TNO Physics and Electronics Laboratory, The Netherlands 
R.J.Dekker@fel.tno.nl 
  
Working Group VII/6 
KEY WORDS: SAR, ERS, Urban Objects, Change Detection, Segmentation, Classification, Interferometry. 
ABSTRACT 
Information on human development and database building are crucial for providing the basic needs and infrastructure in 
fast growing urban areas, as in developing countries. To obtain this information for the area of Dar es Salaam in 
Tanzania, the application of ERS Synthetic Aperture Radar (SAR) remote sensing data was investigated. The objectives 
of this study were (1) the detection of new human settlements on the outskirts of the city, (2) the segmentation and 
classification of land use, and (3) the generation of a Digital Elevation Model (DEM) for storm water flow studies, 
using interferometry. Change detection was done by subtracting two images of 1996 and 1998, and reduce the speckle 
with a special filter. The result shows a high correlation with the ground truth. The fact that most houses contain 
corrugated-iron roofs contributes to this. The segmentation of land use was done on an image pair of 1997, using the 
intensity and textural information. The result is accurate enough for larger scale interpretation. A DEM was also 
generated from the 1997 image pair. Unfortunately its accuracy was insufficient for the application of storm water flow 
studies, due to decorrelation effects. A coherence image was generated from this data set to examine its application for 
land use classification in this area. The overall conclusion of this study 1s that ERS SAR imagery, in combination with 
the techniques presented, has a large potential for the monitoring of urban growth in developing countries. 
1 INTRODUCTION 
In developing countries a number of urban areas is growing fast and mainly unplanned. A problem related to this, is that 
basic needs and infrastructure (health care, water, sewerage, drainage, etc.) are often poor or not available at all. 
Another problem are land use conflicts. An example of an area where the authorities are dealing with such problems is 
Dar es Salaam in Tanzania. Information on human development and database building are crucial in solving the 
problems, but are often a major obstacle (Sliuzas 1997). The most important remote sensing information sources until 
now were SPOT satellite imagery and aerial photography, both optical. To investigate the potential of satellite 
microwave Synthetic Aperture Radar (SAR) imagery for this purpose, ERS SAR data was studied. 
A SAR is an active instrument, and is therefore independent on sunlight and the time of day or night. Because it 
operates in the microwave region, it is also independent on atmospheric circumstances like cloud cover and rain. This 
can be very useful in tropical regions. It is a side-looking instrument resulting in slant (oblique) imagery. Radar 
backscatter has two quantities, an amplitude or intensity, and a phase. A SAR image normally shows the intensity. The 
phase is related to the distance between the illuminated object and the radar. In normal radar imagery it is of less 
importance, but in interferometry it is the key quantity. 
A disadvantage of SAR is that it contains speckle noise, a consequence of the coherent nature of SAR signals. Speckle- 
noise can hamper the interpretation and information extraction, but various filter methods have been developed to 
reduce this problem. One of them is multi looking by averaging adjacent samples. A measure for the quantity of speckle 
is the number of looks. The larger the number of looks, the less the amount of speckle. Another speckle reduction 
method is adaptive filtering. Radar backscatter from urban areas can be strongly dependent on the view angle and the 
nature of the illuminated objects and surfaces (Henderson 1995). 
The objectives of this study were (1) the detection of new human settlements on the outskirts of the city, (2) the 
segmentation and classification of land use, and (3) the generation of a Digital Elevation Model (DEM) for storm water 
flow studies. A DEM can be generated by interferometry. In the next section the ERS SAR and data set are discussed. 
Section 3, 4, and 5 are subsequently on change detection, segmentation and classification, and interferometry. In section 
6 conclusions and recommendations are given. 
  
62 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B1. Amsterdam 2000. 
The 
(19 
of I 
(1) 
(2) 
(3) 
(4) 
The 
sho 
Che 
tabl 
that 
pre: 
196 
ER 
onl 
pha 
nun 
adj: 
dis: 
suit 
the 
cha 
Ta 
To 
sett 
pre 
in ' 
sm: 
bec