programmed by combining domain specific knowledge of landscape planning process 
with map-based visibility analysis. As showed in Fig.3, the analysis model 1 
consists of visual field analysis from single and multiple view 
point (s), hierarchical visual field analysis, front-view analysis and visual 
corridor analysis from a straight line or a curve line, perspective view from a 
view point, occluded landscape analysis, simulation of terrain landscape 
statistics. Since DAM was used as primary spatial data, the analysis model was 
implemented in the raster GIS environment. 
4). Urban flooding analysis model 
The spatial process of the submergence flooding in urban area was modelled by 
combing DTM based analysis and hydrological behavior of flooding([Xiang, 1993]. 
Since urban flooding is a distributed phenomena over the watershed and the 
submerged area, an interactive method was proposed for determing the flooding 
height above the submerged area according to the total quantity of flooding 
from the breach of the riverbank. The propagation of flooding over the submerged 
area was simulated with an ‘inflation’ algorithm and a connected graph was 
designed for recording the spreading path during the propagation. 
maps, but fail to eventually represent much of the evidence available about the 
distribution of spatial entities and fail to support efficiently spatial 
analysis and spatial decision making [Lee, 1990] | 
4. Summary 
In spite of the efforts devoted to the develop of urban GIS's spatial analysis 
models, significant problems still exist, One major problem arises from the 
fact that it is often not possible to define the required precise mathematical 
spatial models due to an adequate theoretical understanding of the spatial 
reality. In addition, it is also not easy to verify and validate the spatial 
model. 
Other problems come from the GIS side or from the coupling techniques. Firstly, 
the so-called 'cartographic data structures’ or the discrete 'tessellation ^of 
current commercial GIS are good at encoding features on maps but fail to 
eventually represent much of the evidence available about the spatial reality. 
There is sometimes the mismatch between spatial reality, the forms of 
discretization used to collect and store data about continuous phenomena and 
the form in which it must be used in the model. Ideally, the imposition of a 
spatial grid whose scale is determined by the spatial scale of the processes 
under study. However, some critical processes operate on many different scales 
in time and space and there may be scale thresholds at which critical processes 
change [Kemp, 1992]. In addition, urban spatial analysis may need more complex 
spatial data models, e.g., a flooding simulation model needs both the vector 
168 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996