Search

The choice of a sampling interval in hydrometry is constrained by two extremes, redundant information from sequentially correlated observations resulting from sampling too often, and lost information because of too infrequent sampling. The variability of the sample variance is used to obtain a sampling interval at some level of information redundance. This statistic is suggested as an alternative measure of information content to the one based on the variance of the sample mean. Time series of stream discharge and temperature measurements are used to illustrate application of the theory. [ABSTRACT FROM AUTHOR]