You are here:
STREAMBED PARTICLE SIZE FROM PEBBLE COUNTS USING VISUALLY ESTIMATED SIZE CLSASES: JUNK OR USEFUL DATA?
Citation:
Faustini, J. M. AND P R. Kaufmann. STREAMBED PARTICLE SIZE FROM PEBBLE COUNTS USING VISUALLY ESTIMATED SIZE CLSASES: JUNK OR USEFUL DATA? Presented at Fall Meeting of the American Geophysical Union, San Francisco, California, Dec. 13-17, 2004.
Description:
In large-scale studies, it is often neither feasible nor necessary to obtain the large samples of 400 particles advocated by many geomorphologists to adequately quantify streambed surface particle-size distributions. Synoptic surveys such as U.S. Environmental Protection Agency's Environmental Monitoring and Assessment Program (EMAP) seek to assess the status and trends of various ecological indicators within populations of streams in regions ranging from a single large watershed to the entire United States. To sample an adequate number of streams, EMAP surveys use a streamlined pebble count procedure with visual size class estimates for 105 particles. To assess the practical applications and limitations of such data, we quantified the precision and sources of error in several measures of streambed particle size distribution from over 800 stream sites in 12 western states surveyed between 2000 and 2003, including geometric mean particle size, Dgm (a proxy for median particle size, D50, that can be computed from size-class data), and the percentage of the streambed surface covered by fine sediments (<2 mm), Psf. Sampling precision was quantified using data from randomly selected sites revisited during the same field season (56 same-year revisits total). We examined three measures of precision: the root mean square error (RMSE) of repeat measurements, the coefficient of variation (CV=RMSE/mean), and a "signal-to-noise" (S:N) ratio defined as the ratio of between-site variance to repeat visit variance. RMSE was 0.19 for log Dgm (equivalent to a 1.55-fold error in Dgm) and 5.3% for Psf. The CV values for the same metrics were 28% and 15%, respectively, and were between 12% and 30% for each metric in 5 of 6 geographic sub-regions (N = 32 to 442 sites per region). For both Dgm and Psf the S:N ratio was >30 for the whole dataset, >12 for 5 of 6 sub-regions and >8 for all sub-regions, indicating adequate precision to characterize within-region variability, including variations that might be associated with land cover alterations or other anthropogenic landscape disturbances.