Identifying Sub-environments in a Tidal Flat: An MDS Approach

Ripple marks formed in different sub-environments of a tidal flat are considered helpful in distinguishing such sub-environments from one another owing to the predominance of specific agencies that generate and stabilize these ripples. For example, ripple stabilized in tidal channels are dominated by tidal currents and those in estuarine settings are both wave and current dominated. In order to evaluate the possibility of using ripple morphology in distinguishing between such sub-environments, data on ripple index (RI) from an estuary and tidal channels were analyzed. These data were collected from the east coast of India that provides examples of open-coast tidal flats at Chandipur and 50 km north-east at Digha. It was found that they have overlapping morphological patterns as delineated from RI vs. the percentage fraction for each case of estuarine and tidal channels. So, while RI can broadly define environments, e.g., aeolian and water borne ripples, when it comes to separating wave from current ripples, a fair overlap is seen in tidal flat regions. On the other hand, grain-size study alone is insufficient for the purpose of delineating differences between such sub-environments. This is because it is seen in our analyses that while most estuarine sediments are finer than 3 on the Φ-scale and tidal channel sediments are coarser, much better sorted as well as mostly positively skewed, these values overlap such that a clear distinction cannot be achieved. Therefore, in order to distinguish between such sub-environments, we took into account both grain-size parameters associated with ripple marks and the corresponding RI values. A relative “distance” was calculated between pairs of samples based on RI, mean, sorting, skewness and kurtosis. In other words, this “distance” is a way to determine how similar two samples are in terms of their respective RI and grain-size parameters. A dissimilarity matrix was constructed which, in turn, was translated into a configuration of points in the Euclidean Space via Multidimensional Scaling (MDS). Four different “types” of measure of this “distance” were considered: Euclidean, Standard Euclidean, City Block and Minkowski (P=3). It was observed that the MDS plot based on City Block distance generated two distinct clusters of points for samples collected from estuary and tidal channels. A way forward is to employ Random Forest Analysis and test whether RI in conjunction with grain-size parameters may be used for classifying modern sediments or even rock samples representing paleo-environments as belonging to either estuarine or tidal channel setting.