This is a new river network data set for Liberia and has been developed independently. It is available in two forms: as an public domain data set at 1:200,000 scale and as a licensed product at 1:25,000 scale.
To identify the river network SRTM data were used to calculate runoff. In turn this was used to isolate streamlines that roughly concurred with the perennial rivers from the Liberia Institute for Statistics and Geographic Information Service (LISGIS). Using this method means that stretches of watercourse are identified which have a comparable minimum volume of stream flow, enabling an objective like-for-like comparison between rivers (assuming uniform rainfall). To roughly match the LISGIS map it is necessary to include a broad range of runoff, including major rivers and more minor creeks.
The LISGIS data set includes feature attributes perennial channelised rivers and unclassified. The SRTM derived data corresponds roughly to the perennial channelised rivers class but does not extend so far up into the catchments.
Like the LISGIS data set the SRTM derived data are not perfect and suffer from a general lack of locational accuracy depending on the terrain morphology, with the locational accuracy declining in flatter areas. This is also related to the somewhat coarse resolution of the data. It was therefore necessary to visually correct each of the SRTM rivers to match Landsat imagery at a scale of 1:10,000. This approach gives an indication of both runoff in the river and a good estimate of the river location at 1:25,000 scale. These features were given a confidence attribute of 1. In the upper reaches of the catchments the rivers / creeks become too narrow to be readily visible on Landsat imagery and may not be easily visible on high resolution GoogleEarth imagery either. In these cases the rivers / creeks were not corrected but were left as generated from the SRTM and given a confidence attribute of 2.
A set of layers was then generated at different scales by generalising the data appropriately to each scale, followed by feature smoothing.
The 1:400,000 data are free to download.
The 1:25,000, together with the river catchments layer, can be purchased from the data store.