Assessing the sediment volume contribution to scattering: bulk density fluctuations

The accuracy and resolution of sediment bulk density measurements are examined. Bulk density from cores is traditionally measured by weight loss of extruded and sectioned 2-cm-thick sediment disks. The actual bulk density is thus an average value for the volume of a 2-cm-long-x-cross-sectional-area sediment core disk. Thus, the values of the measurements as well as the estimates of the correlation lengths may be a function of the disk thickness (sampling interval) and estimation of correlation lengths may be distorted. From a bulk density power spectrum and correlation length estimated from existing data, Monte Carlo realizations of the density were obtained for the sediment volume. From such realizations, we re-applied the same procedures used in the actual core analysis on the simulated cores to obtain a "virtual" bulk density profile and then re-estimated the power spectrum. Because actual density is known from the original data in the simulation, the difference between the parameters used to generate the simulation and the "virtual" parameters is a measure of the distortion. As a result of numerous simulations run to achieve a robust estimate of "virtual" bulk density, we show that laboratory procedures bias (lower) the actual variance of the parameters by averaging or smoothing. In related work, however, the first-order autoregressive approach for estimating correlation length from contiguous disks indicates a bias toward a higher value of the parameter than is appropriate.