SEARCH

SEARCH BY CITATION

Keywords:

  • digital elevation models;
  • change detection;
  • erosion;
  • deposition;
  • gravel-bed rivers;
  • braided rivers;
  • error assessment;
  • digital photogrammetry;
  • laser altimetry

Abstract

System-scale detection of erosion and deposition is crucial in order to assess the transferability of findings from scaled laboratory and small field studies to larger spatial scales. Increasingly, synoptic remote sensing has the potential to provide the necessary data. In this paper, we develop a methodology for channel change detection, coupled to the use of synoptic remote sensing, for erosion and deposition estimation, and apply it to a wide, braided, gravel-bed river. This is based upon construction of digital elevation models (DEMs) using digital photogrammetry, laser altimetry and image processing. DEMs of difference were constructed by subtracting DEM pairs, and a method for propagating error into the DEMs of difference was used under the assumption that each elevation in each surface contains error that is random, independent and Gaussian. Data were acquired for the braided Waimakariri River, South Island, New Zealand. The DEMs had a 1·0 m pixel resolution and covered an area of riverbed that is more than 1 km wide and 3·3 km long. Application of the method showed the need to use survey-specific estimates of point precision, as project design and manufacturer estimates of precision overestimate a priori point quality. This finding aside, the analysis showed that even after propagation of error it was possible to obtain high quality DEMs of difference for process estimation, over a spatial scale that has not previously been achieved. In particular, there was no difference in the ability to detect erosion and deposition. The estimates of volumes of change, despite being downgraded as compared with traditional cross-section survey in terms of point precision, produced more reliable erosion and deposition estimates as a result of the large improvement in spatial density that synoptic methods provide. Copyright © 2003 John Wiley & Sons, Ltd.