Adaptive Cluster Sampling for a Temporal-Scale Population

Adaptive cluster sampling (ACS) is appropriate for rare clustered populations with localization tendencies. Up to now, it has been used exclusively for investigating spatial-scale problems rather than temporal-scale such as this study is dealing with, i.e. sediment transport in rivers. Suspended sediment load is carried mostly during relatively short periods coincide with high flows otherwise negligible. In ACS, more samples from critical river stages can be taken with respect to the aggregation tendencies of sediment loads during transport; thus increasing the level of representativeness of samples. Adoption of ACS to this new area needs further verification and adaptation such as definition of the sampling unit, population frame, neighborhood relation, and threshold. In this study, several scenarios were defined for the purpose of evaluating the ACS in sediment estimation. Numerous sample sets were taken from intensive discharge-load records of Sg. Pangsun River, Malaysia. These sample sets are different with respect to initial sample size, neighborhood relation, and discharge threshold. Total suspended sediment loads were then estimated using modified Horvitz-Thompson method. The comparison made between the symmetric neighborhood relation and the forward method suggested in this study showed that the latter could be used instead of the former in sediment studies without losing the accuracy. The findings also suggested the flow duration curve is a useful tool for ranking initial samples in order to determine an optimum discharge threshold.