Particular attention is paid to areal sampling with clay and with wax. The areal- ... The size distribution of the bottom layer is appropriate for identifying the parent ... C. Kellerhals and Bray (1971) obtained (1) by using a voidless cube model .... that contained finer sizes, the smaller particles were either truncated or ignored ...
PROPERTIES O F VARIOUS SEDIMENT SAMPLING PROCEDURES By Panayiotis Diplas, 1 Member, ASCE and Jon B. Fripp 2 ABSTRACT: Several issues regarding the sampling and analysis of bed material from gravel bed streams are addressed in this paper. The need to convert surface samples into volumetric equivalents, and methods to do so, are explained. It is found that most surface samples are unique and, thus, are not directly comparable. Particular attention is paid to areal sampling with clay and with wax. The arealto-volumetric conversion of samples removed from different sediment structures is also considered. Criteria are proposed for determining the minimum dimensions of unbiased volumetric samples. These criteria can be used to determine the minimum area of an areal sample. The minimum depth of a volumetric sample is shown to be larger than the depth of most surface layers. Thus, a surface layer can not be sampled volumetrically. A technique is presented for sampling a wide size range of underwater surface particles. This technique has been tested in the field and found to provide good results. INTRODUCTION
The characteristics of a given stream are linked to the material that comprises its channel bed. The composition and structure of the bed material is as good an indicator of current river characteristics as it is a predictor of future behavior and possible responses of the entire river system. Research into the characteristics of gravel-bed waterways thus relies upon the techniques used to quantify the material found on the channel surface. Typically, a vertical stratification by particle size can be recognized in the channel bed material of gravel streams. This is evidenced by a surface layer, referred to as either the pavement or the armor layer, that is usually noticeably coarser than the subsurface material (Parker et al. 1982). The following layer, called subpavement, often possesses higher amounts of fine sediment compared with the rest of the bed material. The material in the third, or bottom, layer, is similar in size to that of the subpavement but not as rich in fines (Diplas 1991). The statistical properties of the pavement material are important for determining the roughness and stability of the channel bed. The composition of the subpavement material, especially its fines content, is crucial for evaluating the suitability of the stream for spawning. The size distribution of the bottom layer is appropriate for identifying the parent bed material and its usefulness for gravel mining. The thickness of each of the top two layers is close to the size of the largest grain, while the third layer does not have a predetermined thickness. The material within each layer should be sampled and analyzed separately. Volumetric, or bulk, sampling is considered to be the standard sampling procedure because it results in unbiased samples. This procedure involves removing a predetermined volume of material large enough to be inde'Asst. Prof., Dept. of Civ. Engrg., Virginia Polytechnic Inst, and State Univ., Blacksburg, VA 24061. 2 Grad. Student, Dept. of Civ. Engrg., Virginia Polytechnic Inst, and State Univ., Blacksburg, VA. Note. Discussion open until December 1, 1992. To extend the closing date one month, a written request must be filed with the ASCE Manager of Journals. The manuscript for this paper was submitted for review and possible publication on July 3, 1991. This paper is part of the Journal of Hydraulic Engineering, Vol. 118, No. 7, July, 1992. ©ASCE, ISSN 0733-9429/92/0007-0955/$1.00 + $.15 per page. Paper No. 2223. 955
pendent of the size of the grains that comprise the sample. The material is then subjected to sieve analysis. The result of this analysis is interpreted as a grain-size frequency distribution by weight. It is difficult to characterize bed material of gravel-bed streams because the top two bed layers cannot be sampled volumetrically. Instead, surface-oriented sampling techniques should be used. Reliable methods must be developed for converting a surface sample into an equivalent volumetric sample. The properties of wax and clay sampling methods are examined in this paper. Ways of appropriately using areal sampling and of interpreting the results are described. The minimum dimensions required for a sample to be volumetric are determined. A new hybrid approach for sampling the whole size range of stream bed material is presented. Laboratory and field data are used, to test the results. MODELING SAMPLING OF B E D MATERIAL IN GRAVEL STREAMS
Kellerhals and Bray (1971) demonstrated that areal samples tend to be biased in favor of the coarser grains. To remedy this problem they proposed the following formula to convert an areal sample to an equivalent volumetric sample: p(V - W), = Cp(A - W),Df
(1)
where p(V - W), and p(A - W)( = the percentages of a frequency by weight distribution obtained by volumetric and area sampling, respectively; D, = the geometric mean diameter of the grain size interval i; and C = a proportionality constant that is unique for each areal sample. C is obtained by setting the sum of Cp(A - WtDf for all i's equal to 1 and solving for C. Kellerhals and Bray (1971) obtained (1) by using a voidless cube model to simulate an idealized sediment deposit. Based on this model, they conclude that the exponent, x, is equal to - 1 for all areal sampling techniques. The form of the equivalence criterion expressed in (1) is generally accepted as correct. However, several experimental studies (Anastasi 1983; Proffitt 1980) disputed the x = - 1 exponent for converting a wax sample into an equivalent volumetric sample. These studies show that the x = - 1 exponent overcompensates for the original bias of the wax sample, resulting in an estimated volumetric sample that is finer than actually present. An exponent between - 0 . 4 and - 0 . 5 , with an average of -0.47, has been indicated by experiments performed by Proffitt (1980), Anastasi (1983), and Diplas and Sutherland (1988) for conversion, into their volumetric equivalents, of sediment samples removed by wax. In an attempt to resolve these discrepancies, Diplas and Sutherland (1988) refined the original cube model to account for a 33% porosity. This modified cube model results in a conversion exponent of -0.42 for void-dependent areal sampling techniques, such as wax, and - 1 for sampling techniques that have no void dependence, such as clay. The Kellerhals and Bray (1971) and Diplas and Sutherland (1988) models provide identical results for the other surface sampling techniques. EXPERIMENTAL METHODS AND PROCEDURES
The accuracy of the exponent used in the conversion formula is tested in this study by removing areal samples from unstratified gravel mixtures of known size distributions. The size distribution obtained from the areal sam956
pie is then compared with the exact volumetric distribution, and the conversion exponent x is determined by fitting (1) to the data through a loglog least squares approximation. For the gravel material sampled areally by wax, the values of the median grain size, Ds0, range from 0.8 mm to 15 mm, and the values for the geometric standard deviation,
