Summary

Sediment and nutrients play a critical role in fluvial ecosystems. Given growing concern about the environment, it seems likely that scientists, government regulators and resource managers will demand better information about the movement of sediment and nutrients through the environment.

These demands are being addressed. Researchers have begun to construct sound statistical foundations for load estimation, and two recent developments show particular promise:

1. Rating curve estimators: Retransformation bias is now widely understood, and methods for eliminating it are available; the variance of long-term load estimators has been worked out in closed form; ``less-than'' data values can be incorporated into the analysis. These improvements to the rating curve can be implemented without altering current sampling practices. Split-sample studies suggest that bias-corrected rating curve estimators are both accurate and precise for estimating nutrient loads from mid-sized and larger basins.

2. Stratified Random Sampling: Thomas's stratified sampling methods constitute an important innovation. They provide inherently unbiased estimates with known variance. Furthermore, given a crude model of sediment or nutrient dynamics, one can design a nearly-optimal sampling program. For small catchments, stratified random sampling may provide the only alternative to nearly-continuous monitoring.