163 
3.3 ESTIMATION OF FLOW ACCUMULATION BASED ON THE DRAINAGE DIRECTIONS. 
The estimation of the flow accumulation is computed by tracing the waterway upstream from an 
outlet (or sink). When the cell with the minimum elevation in the DEM is found, all cells that drain to 
this cell will contribute to the flow. For an individual cell, there is: 
F out ~ F m + Fi oca i (10) 
where F out denotes the flow output from the cell; F in denotes the flow received by the cell; and Fi oca i 
denotes the flow produced in the cell which is set to 1. 
The F ou , is then distributed to the cell(s) which receive the flow from the centre cell according to its 
drainage distribution. All cells in the drainage area contribute in the same way until the whole drainage 
area to the starting cell (the outlet or sink) is examined. Then the cell with the second lowest elevation 
value in the DEM is examined in the same way, and so on until all cells in the DEM has got a flow- 
accumulation value. 
4.0 RESULTS AND DISCUSSION 
The proposed algorithms have been tested in comparison with some commonly-used algorithms. A 
DEM was acquired in the area of Bredbo. New South Wales by digitising a detailed 1:7,500 contour map 
which was then converted into a grid-based DEM, using ARC/INFO Triangulated Irregular Network 
(TIN) and linear interpolation functions. This results in a DEM of 745 columns and 687 rows with a 5 x 
5 metre grid cell size. 
The proposed algorithms ware tested against the commonly-used ‘eight direction' algorithm 
represented by ARC/INFO hydrological modelling functions, namely flow direction and 
flow accumulation, to produce flow accumulation grid from the DEM. Figure 12 shows the comparison 
between the results. As showm in Figure 12, the flow accumulation produced by the ‘eight direction' 
algorithms has shown a clear weakness for only allowing one of the eight possible flow directions for 
each cell. As a consequence of this limitation, the trend of distribution of flow accumulation in relation 
to land surface was not clearly simulated in the result (Figure 12c). The proposed algorithms, on the 
other hand, have demonstrated its ability to produce a more realistic simulation result (Figure 12a and 
12b) by allowing more complicated scenarios in its simulation of flow direction.