The study of the morphological characteristics of rivers and of their degree
morphological alterations, is a basis for a proper management of mountain watershed:
the availability of detailed topographic data is a key tool. The evaluation of channel
geometry variability, determined by hydrodynamic and geomorphological processes, is
usually gathered through field surveys, or through visual interpretations of digital
orthophotos. However, the topographic data obtained through visual interpretation are
not sufficiently accurate to allow the identification and the correct mapping of channel
geometries. On the other hand, even if they provide more reliable data, field surveys
require considerable time and financial resources, and they are often challenged by the
inaccessibility of the areas under analysis. It is therefore strategic to adopt new and
more accurate methods to estimate channel geometries, based on the availability of
high-resolution data, such as the one derived from airborne laser scanner (LiDAR).
LiDAR technology enables the acquisition of high resolution topographic data over
large areas, with vertical and horizontal accuracy of a few centimeters (10-20 cm for the
vertical component and 0.5-2 m for horizontal), contributing to a better representation
the Earth's surface at more affordable costs. In mountain areas many studies have
explored the potential of LiDAR DTM for the proper characterization of the network
and the objective of this paper is to highlight their potential in the automatic
determination of values representative of bankfull widths.
The analysis is based on a topographic index (Elevation Percentile) used to measure the
variability of the elevation from 1 m DTM resolution. This index is derived by
calculating within a moving window the number of cells with elevation higher than the
central pixel. This number is then normalized to the extension of the moving window:
channelized areas have an EP value greater than convex areas. Applying a statistical
threshold to the EP, it is possible to obtain a Boolean map of potential river bed. The
geometries derived from this map are approximated due to the resolution of the DTM,
but it is possible to consider the topographic representation of the banks and to
approximate the bankfull width. In an automatic manner, moving downstream along the
thalweg, it is possible to estimate perpendicularly to the flow directions, the bankfull
width at each point of the network, with the above-mentioned limit of the DTM
resolution. Automatically derived widths show a good agreement with those detected in
the field, with low values of RMSE, and the range of the estimated values is compatible
with the surveyed ones.
