Method of watershed declination using GIS technologies


                                                                                                             Ing. Jean Albert Doumit

                                                                                                                                  KUBAN STATE UNIVERSITY-RUSSIA

                                                                                                                     Geoinformatics departments


Watersheds are important geomorphologic features, which play an important role in hydrological GIS applications. Watersheds can be re­garded as the lines that separate the area where water drains to different locations. The areas that are enclosed by the wa­tersheds are precisely the regions where water drains to the same place, and are conventionally called basins or valleys.

One of the major categories of approaches to extract wa­tersheds are the ridge detectors, which were first proposed in the early part of the 19th century [R. Rothe(1915)] J. J. Koenderink and A. J. van Doorn(1994)].

Another theory was proposed by Maxwell, Jordan, and Cayley [J. J. Koenderink and A. J. van Doorn(1994) , J. H. Rieger(1997)]. It is based on the observation that for generic surfaces there is a unique slope line through every non-critical point of the surface. However, it suffers from some poor implementation choices that lead to inaccurate results and some surprising cases, in which, for example, slope lines can cross each other.

A third way to extract watersheds from the image is to use the fact that water will accumulate at the minima of the landscape. This means that each minimum in the image defines a valley or water catchment basin. Watersheds are the boundaries between different basins. This can be implemented by flooding the landscape from the minima [L. Vincent and P. Soille(1991)].

Since all of the approaches to extract watersheds return the result only with pixel resolution, a pixel accurate wa­tershed extraction algorithm is desirable. This means that either the definition of Maxwell, Jordan and Cayley, or the definition of Rothe must be used.

GIS and digital elevation models (DEM) can be used to perform watershed delineation; watersheds can be delineated quickly and with consistent time response, regardless of the DEM resolution. 

The most straightforward GIS technique for watershed delineation consists of the following steps [ESRI, 1997; Olivera and Maidment, 1999]:

·          Determine flow direction grid.

·          Determine flow accumulation grid.

·          Specify a "stream" threshold on the flow accumulation grid. This operation will identify all the cells in the flow accumulation grid that are greater than the provided threshold.

A new grid is formed from those cells ("stream" grid). This grid will be an indication of the drainage network. A high threshold area value will create a drainage network with the main streams(less dense network and less internal sub watersheds). A low threshold area value will also create drainage network elements where flow tends to accumulate (dense network and more internal watersheds). In this respect, several authors have pointed out that a variable threshold area should be considered for areas with different relief characteristics [Da Ros, D and M Borga. 1997; Gandolfi, C and G B Bischetti. 1997]

·          Stream grid is converted into stream segments

·          Watersheds (in grid format).

·          Watershed and stream grids are vectorized to produce watershed and stream polygon s and polylines.

As introduced above, the delineation of the drainage network of an area from a DEM requires the establishment of a threshold area. In the present research, several comparisons between a drainage networks obtained from DEM and drainage networks obtained by using different threshold area values were made, in order to set the threshold area.

The considered threshold area values were between 5000 cells, 10000 cells, 15000 cells and 20000 cells, tab 1.




Basins quantity










Table 1: the value of threshold and the quantity of basin got basing to the threshold value.


The idea behind this operation was to obtain different equations to estimate the best fit threshold area, with the only input of easy derived topographical variables.

As an example, figure 1 shows this situation after applying different threshold area values to the flow accumulation grid of the Lebanese territory. Table 1 shows the threshold area values that produced the most faithfully representation of the drainage networks of the Lebanese territories.


Fig 1:  different threshold value of the Lebanese watershed.


The present research showed that very different drainage network representations can be produced by using different threshold area values, the most reliable threshold area, according to the purposes of the user.

As the threshold value is increased, the density of the drainage networks decreases.

The threshold used to delineate the network may vary from one watershed to another since it is dependent on factors such as land use and soil type. Arbitrary methods of threshold value selection without considering factors such as local terrain slope, soil properties, geology, infiltration capacity, surface cover and climatic conditions can lead to erroneous networks [Garbrecht and Martz, 1993].

During the experiments, it was always necessary to edit the delineated watershed. This necessity arises, due to errors in height that are always found in a DEM.

The geographic information system and the digital elevation model reach and advance step of geomorphologic and hydrological analysis.