2.4 Biophysical suitability: multivariate analysis
The first step of our approach is the selection of biophysically suitable location in the province of study. This step is necessary to identify a first set of location with the necessary biophysical condition for the large area of the ten communities. Moreover, being mobile pastoralists, the communities could make good use of water points located even tens of kilometres far from the core of their villages. Selecting suitable location in such a large area would be impossible with a pure participatory work. We then use Multi-criteria Decision Analysis (MCDA) to select a first set of points for the participatory phase. MCDA uses a set of parameters to discriminate between suitable and unsuitable conditions leading to a certain result. It is a consolidated approach in siting suitability analysis of water harvesting structures (Al-Adamat et al., 2010; Forzieri et al., 2008; Grum et al., 2016). In a recent review of siting methods, Ammar et al. (2016) reports MCDA as one of the most reliable approaches, especially when combined with spatial analysis tools, since it allows to analyze several spatial indicators for large areas in a quick and inexpensive way.
Given the uncertainty in selecting the most adequate parameters for the biophysical suitability of sand dams (Ngugi et al., 2020) and the mentioned lack of reliable and high resolution data, we adopted a conservative approach using a simple set of parameters representing the minimum requirement of measurable variables for the construction of sand dams, including the slope, the salinity and the stream order. The purpose of the analysis is therefore to narrow down the potential number of water stream sections to be inspected during the second phase in the field.
The analysis included the delineation of the stream and the calculation of the stream order (or Strahler order), and the subsequent selection of the points along the stream with appropriate conditions of stream order, slope and salinity. Hereinafter we report the analytical steps with some explanatory figures.
To select the final set of water stream sections we have first produced the hydrographic network (i.e. water streams) starting from the Digital Elevation Model (DEM), using the GIS functions slope, aspect and exposition applied at pixel level from the DEM grid. From the slope and exposition layers, we then calculated the flow accumulation, which is the number of upstream cells draining into each cell of the grid. From the flow accumulation we then reconstructed the stream network by selecting the threshold of 1000 cells from the flow accumulation layer. After reconstructing the river layer, we broke the stream lines into sections 30m apart, technically by converting the stream lines into points. We yield a total of 1813428 points, which represent all the stream sections in Namibe, then we narrowed down the selection of suitable point by excluding the points located in areas with slope higher than 2 degrees, stream order higher than 2 and soil salinity higher than 40Hz. The stream order is a proxy for stream dimensions, since sand dams are most suited for river sections below 25m (Beswetherick et al., 2018). Finally, soil electrical conductivity below 40Hz is a common acceptable salinity value for drinking use (following the FAO guidelines for maximum electrical conductivity values:
http://www.fao.org/3/x5871e/x5871e04.htm).