DTM and interpolations in GRASS: application to a braided river

The present work deals with the problems related to the generation of Digital Terrain Model (DTM) of braided rivers by means of the open source GIS GRASS. In particular, we investigated how to optimize the terrain survey, that is how to reduce the density of topographic points that we have to survey and which is their best distribution, related to the employed interpolation method and with respect to the wanted precisions and to the surface description necessary for the specific application. Moreover, we dealt with the delicate problem of calibration of interpolation parameters. A braided river is characterized by the existence of more than one channel contributing to flow and sediment transport on the floodplain, giving rise to frequent bifurcations followed by confluences. The channels are in continuous motion due to bed and bank scour during floods of even moderate intensity, hence braided rivers pose problems of river regulation and management. To investigate the processes that control the morphodynamics of braided networks, fluvial engineers preliminary need detailed topographic data of both planimetry and altimetry, but they face with onerous field data collections needs. The present work originates from the necessity to monitor quickly and economically a short reach of a braided river, so to get information sufficiently detailed and frequent in time. The limited dimension of the area to monitor allows to survey the territory through GPS or traditional ground survey, hence more economically than through aerial photogrammetry and laser technique. Moreover, such ground techniques allows to survey also the wet portions of the river, that may occupy more than 50% of the floodplain. We proceeded as follows: we created a synthetic DTM representative of a straight, trapezoidal channel in a floodplain, that bifurcates in two channel that themselves rejoin downstream in a single one. Such synthetic DTM has the advantage to be characterized by a simpler geometry that in nature, but, at the same time, it reproduces some important features of the real morphology. From such DTM we extracted different data structures (transversal cross section, slope changes, scattered points, etc.) so to simulate the ground survey; then we recreated the DTM by GRASS, interpolating such data varying their density and distribution. The interpolation methods investigated are: the inverse distance weighted (IDW), the spline with tension (RST) and kriging (interfacing GRASS with R). We compared the potentialities of different interpolating methods in function of data density and distribution and their …