158
3.0 METHODOLOGY
The proposed algorithms contain functions for estimating drainage distribution and flow
accumulation, and can be presented in the following tasks. The two algorithms are presented as The
aspect based algorithm' and ‘the form based algorithm’.
3.1 THE ASPECT BASED ALGORITHM
3.1.1 Estimation of drainage vectors based on the aspects values.
Skidmore (1989) compared six commonly used methods to generate gradient and aspect from
gridded DEM. He concluded that general linear regression models and the third-order finite difference
methods were more accurate than methods based on maximum gradient drop, maximum gradient rise or
second-ordered finite difference. Since the linear regression models are far more computing intensive, the
third-order finite difference method was chosen for the applications. The first step in the algorithm,
proposed by Horn (1981), is to calculate
least one neighbour, in a 3 x 3 window, which has a lower elevation value than the centre cell are
examined. The remaining cells, such as sinks or cells on flat areas, are flagged and then treated
separately. The aspect estimation of the flagged cells are done by vector addition of the aspect vectors
from the surrounding cells. Firstly, all flagged cells with eight neighbours with defined aspect values, e.g.
single sinks, are examined. The eight vectors are summed and the resulting vector is set to be the aspect
of the flagged cell. Then all flagged cells with seven neighbours with defined aspect values are
examined, and so on, until all the cells in the DEM have got a defined aspect value.
dz
( 1 )
8Z)X
and
dy
dz
( 2 )
where Z[j denotes the centre cell (pixel) of a 3 x 3 moving window located at the zth row andyth
column; DX denotes the distance between points in the horizontal direction; and D Y is the distance in
vertical direction (in our case; DX= D7= cell size).
The aspect (A) is then defined as
dz / dx
( 3 )
Aspect values are calculated for the cells in DEM using the above equation. Only cells with at
