河流砂碎屑统计方法的对比实验研究

碎屑统计是定量研究陆源碎屑沉积物常用的方法，其中最具代表性的有四种:全颗粒法、Glagolev-Chayes法、Gazzi-Dickinson法和全面积法。目前对以上方法的差异性缺乏定量研究，其统计的颗粒数各不相同（从300颗到500颗不等），也缺乏样品分选程度和颗粒含量对统计结果影响的探讨。选取雅鲁藏布江干流的两个河流砂样品，分别用四种方法进行定量颗粒统计，并结合Matlab模拟统计开展实验。将全面积法统计结果作为最能够代表样品真实成分的标准，结果表明，Glagolev-Chayes法与全面积法的统计结果相近，全颗粒法与全面积法的偏差较大；对于分选越差的样品，全颗粒法与全面积法统计结果的差异性越大；统计颗粒数达到384颗时满足95%的置信度；在统计过程中应重点关注含量小于10%的颗粒类型，若岩屑总量小于10%时建议不再进行Lv/Lm/Ls投图。在计算河流通量时建议采用全面积法或Glagolev-Chayes法进行碎屑统计。

Comparative Experimental Research on Detrital Statistics of River Sand

Statistical analysis is the most effective way of determining the component and content of sand or sandstone， and is widely applied in the analysis of terrigenous detritus. Methods include grain counting， the Glagolev-Chayes method， the Gazzi-Dickinson method and the area method. The number of grains to be taken into account in each of these approaches is controversial， and ranges between 300 and 500. The extent to which the statistical result is affected by the sorting degree and grain percentage has not previously been addressed. In this study， two modern river sand samples were collected from the main stream of the Yarlung Tsangpo River for statistical analysis using the four methods， in addition to Matlab simulation statistics. The area method was adopted as the standard which most closely represents the true composition of the sample. It was found that the Glagolev-Chayes method was the most similar to the area method， and the grain-counting method gave the greatest difference. Moreover， poorer sorting of the samples increased the discrepancy between the grain-counting method and the area method， leading to a statistical bias for the components comprising less than 10% of the sample， and the Lv/Lm/Ls diagram is not recommended in this case. The count number is no less than 384 to ensure 95% confidence. Either the area method or the Glagolev-Chayes method is highly recommended rather than the Gazzi-Dickinson method for detrital statistics when calculating river flux.