Memoire Online - Hydrological modeling of the Congo River basin: Asoil-water balance approach

This section presents and describes the findings of the DEM-Hydro processing technique accordingly to their application in hydrological modelling.

The topography is mostly characterised by a more or less flat are in the centre of the study area (Figure 14). This area is called «Central cuvette» and is limited by the Great Rift Valley to the East, mountainous regions in the north-western and south-eastern corner of the study area.

The altitude varies between -99999 and 4657 m with an average of 1886 m. In HYDRO1k DEM, pixels with missing data are assigned a negative value of -99999. Extracting the area covering exclusively the Congo Watershed, the elevation mean is around 238 m aswl with a minimum of 0 m.

Figure 13 Areal distribution at different altitude (The area in a logarithmic scale)

Figure 14 DEM visualization map for Cental Africa. The defined colored polygone delineated the Congo River basin.

Elevation	npix	npixpct	npixcum	npcumpct	Area (Km square)
0-100	46440	0.65	41461805	553	46956
101-200	88017	1.24	47682206	636	88994
201-500	1984879	27.92	342086982	4563	2006916
501-750	1770620	24.91	869454897	11598	1788574
751-1000	901295	12.68	1183992072	15794	911302
1000-1500	2006891	28.23	3176337475	42371	2029172
1501-2000	274325	3.86	3693851852	49274	277371
2001-2500	28808	0.41	3738833270	49874	29128
2501-3000	6081	0.09	3716166768	49572	6149
3001-3500	1168	0.02	2840777253	37895	1181
3501-3999	467	0.01	1866541342	24899	472
4000-4005	n/a	0.00	-	-
4006-4657	114	0.00	607211659	8100	115

PS: npix= number of pixels, npixpct= percentage of number of pixels, Npicum = cumulated percentage of number of pixels. In colone 2, the pixel numbers with -9999 elevation value are ignored.

This step comes after fill-sink step. The filled DEM was then used to find the flow direction map using standard D-8 algorithm (Figure 15). Flow direction is calculated for every central pixel of input blocks of 3 by 3 pixels, each time comparing the value of the central pixel with the value of its 8 neighbors. The steepest slope method was used for this study to find the steepest downhill slope of a central pixel to one of its 8 neighbour pixels and assign to flow directions.

Figure 15 D-8 algorithm: Based on the output Flow direction map, the Flow accumulation operation counts the total number of pixels that will drain into outlets (after ILWIS 3.4 Manual)

The output map shown in Figure 16 contains flow directions grids as N (to the North), NW (to the North West), NE (to the North East), SE (to the South East), S (to the South) and SW (to the South West).

The histograms (Figure 17) indicate that the flow direction algorithm tends to favor the cardinal directions (north, south, east and west) over the diagonal directions (northeast, northwest, southeast and southwest). For the entire dataset (rectangular area) 63 % of grids cells had flown in a cardinal direction as compared to 37 % diagonals. This indicates that the flow direction algorithm used in the model is predisposed in favor of flow through the cardinal directions. The same observation was done in previous study on the basin (Kwabena, 2000).

Table 5 Summarised statistics for the Flow direction grid map in the area of study.

Flow direction Orientation	Number Pixel	% Number pixels	Area (Km2)
E	579477	15	579477
N	536518	14	536518
NE	334119	9	334119
NW	372838	10	372838
S	600426	16	600426
SE	314363	8	314363
SW	382729	10	382729
W	658407	17	658407
Min	314363	8.3	314363
Sum	3778877	100.0	3778877

The flow direction grid developed at the previous step is then used as input data for Flow Accumulation grid calculations. The flow accumulation map contains cumulative hydrologic flow values that represent the number of input pixels which contribute any water to any outlets; the outlets of the largest streams (drain, river) will have the largest values which is 3778906 for the Congo River basin. The grid generated (Figure 18) has a minimum of 1 and a maximum of 3778906 pixels values for computed flow accumulation matrix.

Figure 18 Flow Accumulation map; on top: Entire basin, on bottom: A selected area

The Drainage Network Extraction operation extracts a basic drainage network (raster map). As input it is required the output raster map of the Flow Accumulation operation and a defined threshold value. A threshold value, i.e. a value for the minimum number of pixels that are supposed to drain into a pixel to let this pixel remain as a drainage in the output map; the larger is this value, the fewer drainages will remain in the output map. Depending on the flow accumulation value for a pixel and the threshold value for this pixel, it is decided whether true or false should be assigned to the output pixel. If the flow accumulation value of a pixel exceeds the threshold value, the output pixel value will be true; else, false is assigned. A threshold value of 1000 (number of pixels) is used in this process and 1752 stream segments are identified in the Congo River Basin masked (Figure 19).

Figure 19 Stream network map masked by the boundary of the Congo River Basin 4.3.5 Catchment and Sub-Catchments extraction

During the Catchment extraction operation, 3435 sub-catchments were extracted. Using the Cross operation 1752 sub-catchements only (Figure 20) were selected; each of them corresponding to a single stream segment from the Drainage network ordering operation. This operation delivers an output raster map, an output polygon map and an output attribute table.

The attribute table (appendix 6) andd Table 6 summarises information for each catchment, such as the area, longest flow path, density, and perimeter of catchment, the total upstream area. Figures 21-22 show 5 sub-catchments corresponding to 5 defined outlets namely Sangha, Ubangi, Kasai and Lualaba. The Congo sub-catchment is generated with the residual area.

Figure 21 Merged sub-watershed with stream network and majors outlet of the CRB

Hydrological modeling of the Congo River basin: Asoil-water balance approach

4.3 DEM-Hydro processing output maps