IMIS | Flanders Marine Institute
 

Flanders Marine Institute

Platform for marine research

IMIS

Publications | Institutes | Persons | Datasets | Projects | Maps
[ report an error in this record ]basket (0): add | show Printer-friendly version

Sediment transport paths in the westerschelde: one-dimensional alternative to determine sediment trend
Lucio, P.S.; Dupont, H.S.; Bodevan, E.C. (2004). Sediment transport paths in the westerschelde: one-dimensional alternative to determine sediment trend. J. Coast. Res. 20(3): 771-775
In: Journal of Coastal Research. Coastal Education and Research Foundation: Fort Lauderdale. ISSN 0749-0208, more
Peer reviewed article  

Available in Authors 

Author keywords
    Autocorrelation function; Binomial distribution; Principal components analysis; Sediment trend analysis

Authors  Top 
  • Lucio, P.S.
  • Dupont, H.S.
  • Bodevan, E.C.

Abstract
    The sediment granulometry is a complex feature that requires studies and definition of parameters that go beyond a simple categorisation or classification. The grain size distribution frequencies may be analysed under a statistical approach, to establish parameters concerning the residual sediment transport. In estuaries, the sediment transport paths investigation is important to maintain navigation channels by dredging. Different methods based on grain-size parameters were developed to investigate these transport paths. These methods consider the spatial changes from three grain-size parameters: mean, standard deviation (selection) and skewness. In this work, we propose two alternatives to the one-dimensional (1-D) Sediment Trend Analysis (STA®), introduced by MCLAREN (1981). The first one is based on a non-parametric statistical test called Structural Exact Test (SET), which consider that each line memory has size not greater than one according to LUCIO et al. (1999), where an asymptotic test was adapted for determining the order of a Markov Chain. Thus, observing the validity of all suppositions for the binomial distribution modelling checking the memory by means of the autocorrelation of the sequence or their difference, it is possible to implement the (SET). The second method is based on techniques of multivariate statistics like Principal Components Analysis (PCA) that allow us to associate the three-grain-size statistical parameters to create a characteristic index and concatenate their results to the difference sequences given the Autocorrelation Function (ACF). The ACF computes the autocorrelations of a sequential stochastic process. Autocorrelation is the correlation between observations of a sequence separated by k units. The Partial Autocorrelation Function (PACF) computes the partial autocorrelations of a sequential stochastic process. PACF, like ACF, are correlations between sets of ordered data pairs of a sequence. As with partial correlations in the regression case, partial autocorrelations measure the strength of relationship with other terms being accounted for. Based on the scope of Time Series the partial autocorrelation at a lag of k is the correlation between residuals at time t from an autoregressive model and observations at lag k with terms for all intervening lags present in the autoregressive model. The study area is based on the Westerschelde between Baarland in Holland and Rupelmonde in Belgium, where 867 sediment samples were collected by GeoSea Ltd in 1993, for sediment trend analysis. These samples were used to check our methods: the SET and the PCA, comparing the results with the STA methodology. Our approach seems to be robust and offer a high efficiency results for a low computational cost in comparison with the classical and traditional one-dimensional sediment trend analysis.

All data in IMIS is subject to the VLIZ privacy policy Top | Authors