 |
| Titel |
Collection, processing and error analysis of Terrestrial Laser Scanning data from fluvial gravel surfaces |
| VerfasserIn |
R. Hodge, J. Brasington, K. Richards |
| Konferenz |
EGU General Assembly 2009
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 11 (2009) |
| Datensatznummer |
250025929
|
|
|
|
|
|
| Zusammenfassung |
| The ability to collect 3D elevation data at mm-resolution from in-situ natural surfaces, such
as fluvial and coastal sediments, rock surfaces, soils and dunes, is beneficial for a range of
geomorphological and geological research. From these data the properties of the surface can
be measured, and Digital Terrain Models (DTM) can be constructed. Terrestrial Laser
Scanning (TLS) can collect quickly such 3D data with mm-precision and mm-spacing. This
paper presents a methodology for the collection and processing of such TLS data, and
considers how the errors in this TLS data can be quantified.
TLS has been used to collect elevation data from fluvial gravel surfaces. Data were collected
from areas of approximately 1 m2, with median grain sizes ranging from 18 to 63 mm. Errors
are inherent in such data as a result of the precision of the TLS, and the interaction of factors
including laser footprint, surface topography, surface reflectivity and scanning geometry.
The methodology for the collection and processing of TLS data from complex
surfaces like these fluvial sediments aims to minimise the occurrence of, and remove,
such errors. The methodology incorporates taking scans from multiple scanner
locations, averaging repeat scans, and applying a series of filters to remove erroneous
points.
Analysis of 2.5D DTMs interpolated from the processed data has identified geomorphic
properties of the gravel surfaces, including the distribution of surface elevations, preferential
grain orientation and grain imbrication. However, validation of the data and interpolated
DTMs is limited by the availability of techniques capable of collecting independent elevation
data of comparable quality. Instead, two alternative approaches to data validation are
presented. The first consists of careful internal validation to optimise filter parameter values
during data processing combined with a series of laboratory experiments. In the experiments,
TLS data were collected from a sphere and planes with different reflectivities to
measure the accuracy and precision of TLS data of these geometrically simple
objects.
Whilst this first approach allows the maximum precision of TLS data from complex
surfaces to be estimated, it cannot quantify the distribution of errors within the
TLS data and across the interpolated DTMs. The second approach enables this by
simulating the collection of TLS data from complex surfaces of a known geometry.
This simulated scanning has been verified through systematic comparison with
laboratory TLS data. Two types of surface geometry have been investigated: simulated
regular arrays of uniform spheres used to analyse the effect of sphere size; and
irregular beds of spheres with the same grain size distribution as the fluvial gravels,
which provide a comparable complex geometry to the field sediment surfaces. A
series of simulated scans of these surfaces has enabled the magnitude and spatial
distribution of errors in the interpolated DTMs to be quantified, as well as demonstrating
the utility of the different processing stages in removing errors from TLS data.
As well as demonstrating the application of simulated scanning as a technique to
quantify errors, these results can be used to estimate errors in comparable TLS
data. |
|
|
|
|
|
|