Lebialem Highlands Great Apes Research and Conservation Program, South West Cameroon.

Drainage and Catchments over the Lebialem Highlands region: GIS Methodology employed Produced for the African Conservation Foundation (ACF) By Lauriane Cayet-Boisrobert, MSc. ([email protected]) & Magali Moreau, MSc. and current PhD candidate ([email protected]) Volunteers for ACF, through the UN Online Volunteers Program January 17th, 2007 Note: The analysis was run over a small region, delimited by a polygon which was provided by ACF as a shapefile (rectangle1.shp). The region of interest is located in the area defined by the following coordinates: xmin: 9d41’E; xmax: 10d 6’E; ymin: 5d21’N; ymax: 5d46’N.

1

Selecting appropriate elevation material

Tiles 38_11 and 39_11 were downloaded from the CGIAR website (http://srtm.csi.cgiar.org/). The CGIAR STRM 90m product seemed relevant for the purpose of the exercise as they applied a hole-filling algorithm to provide continuous elevation surfaces. The area of study is located mostly within tile 38_11 and partially within tile 39_11. To obtain the elevation information over the area of study, the two tiles were mosaiced using the mosaic tool in ArcToolBox (Data Management Tools / Raster / Mosaic). The resulting raster was clipped with the polygon of the area of study using the in ArcToolbox. The coordinate system of the rectangle shapefile was changed to WGS 1984 to match the coordinate system of the SRTM data. The final output was called LH_DEM.

2

Modelling watersheds of the study area

The following steps were performed in sequence to model watersheds from the Lebialem Highlands DEM. These were all performed using the Hydrology Toolbox in ArcGIS Spatial Analyst. 2.1

Creating a depressionless DEM

The sinks in the DEM were filled to create a depressionless surface. It is important to have a depressionless DEM for all subsequent hydrological analyses, as areas of internal drainage can cause problems in the watershed delineation process. Running the function, several sinks were detected. These are natural features but need to be filled for the purpose of the exercise. These sinks were filled using the function, with the DEM as input. The output was called Fill_DEM.

Boisrobert-Cayet, L. , Moreau, M.

1

Lebialem Highlands Great Apes Research and Conservation Program, South West Cameroon.

2.2

Defining the stream network

The direction of flow for each cell was calculated using the function, with the depressionless Fill_DEM as input. The flow accumulation was calculated using the function, with the flow direction output obtained in the last step as input. The higher values of flow accumulation indicated the location of the drainage network. The stream network was determined by extracting the cells in the flow accumulation layer which had values below a determined threshold value. This threshold value was selected on the basis of a standard deviation classification of the flow accumulation layer. The resulting raster was then converted to a polyline shapefile, using the ArcToolbox Conversion Tools. 2.3

Building the topology of the stream network

The topology of the stream network was built using the Stream Order function with the Shreve classification scheme. The output was called Streams and topology information is contained in the field of its table of attributes. When taking a closer look at Streams, loops were clearly identified at some of the convergences of two river branches. Flat areas within the DEM cause the production of loops when deriving the drainage. These loops cause erroneous classification of the river according to the Shreve scheme at their exact location. Thus, loops also cause difficulties when deriving the pour points, causing erroneous additional vertices. 2.4. Cleaning the river network Having a clean drainage network derived from the DEM is important for several reasons. First, a clean river network (without any loops) is necessary to derive the pour points. Second, a clean network would provide ACF with a support tool to digitize the stream network, if the need presented itself. A geodatabase was built, and a geometric network of edges of the drainage network was imported. Flow direction information was integrated to the geometric network (a sink information file was made to support building the directionality). Using the utility network extension, the direction of the network was defined. Edges without a direction showed the location of the loops. These tiny edges were then deleted from the Streams file. Since the loops had created errors in the field, standing for the codification of each edge of the river networks, all the erroneous numbers were manually edited (following the Shreve nomenclature). 2.5.

Creating the pour points dataset.

Pour points are necessary to delineate watersheds. The drainage file Streams was used to obtain the pour points dataset by selecting the endpoints of each edge using the tool. However, for an unexplained reason, the tool did not select end points only. Several other edges which were not terminating an edge but were starting one were extracted. To address this issue, the points associated with non-terminating edges were manually deleted. Boisrobert-Cayet, L. , Moreau, M.

2

Lebialem Highlands Great Apes Research and Conservation Program, South West Cameroon.

2.4

Deriving the watersheds

The watersheds of the area were derived using the pour points layer and the flow accumulation layer as inputs. This resulted in a raster which was vectorised to produce the basins dataset, with LH_basins.shp as final output. This dataset does not include any topology information. Some basins were erased because they were at the border of the region, and hence were erroneous. An Aspect raster file was created to support our decision-making.

Boisrobert-Cayet, L. , Moreau, M.

3