地统计法支持的北部湾东部海域沉积物粒径趋势分析

在北部湾东部海域采集表层沉积物71 个， 通过粒度分析获得其粒度参数(平均粒径、分选系数、偏态)， 用克里格插值法将不规则采样站位的粒度参数内插为规则网格分布的相应粒度参数。采用地统计法分析其空间相关性， 计算度量空间相关性范围的参数， 即半方差图中的变程值。结果表明， 使用地统计分析获得的粒度参数变程值物理意义较为明确， 可作为粒径趋势分析模型的特征距离， 其中分选系数变程值作为特征距离的计算结果与前人的海流、沉积物输运信息更为吻合； 这在一定程度上消除了传统方法(试算法或经验估计法) 获取特征 距离可能造成的模型计算误差。采用不同间距插值时得到的粒径趋势矢量具有不同的空间分 辨率， 其中高分辨率的细化图所反映的海底沉积物净输运趋势与余流和环流等所反映的沉积物输运细节特征吻合较好， 低分辨率的概化图可大致反映该区域沉积物的总体输运趋势。地统计分析的结果对未来研究工作中采样间距的选取也具有指导意义。

The Application of Geostatistics to Analysis of Grain Size Trend in the Eastern Beibu Gulf

A total of 71 surficial sediment samples are collected in the eastern Beibu Gulf, and then the moment parameters (i.e. mean size, sorting coefficient and skewness) are obtained by grain size analysis. By using kriging interpolating methods, the regular meshes of grain size parameters are calculated. A geostatistic analysis is applied to study the spatial autocorrelation for these parameters; while the range, a parameter in the semivariogram metering the bound of the spatial autocorrelation, is estimated. It shows that the range for sorting coefficient is physically meaningful, and also the obtained grain size trends are consistent with the annual ocean circulation pattern and sediment transport rates from previous studies. Thus, the range from the sorting coefficient variogram can be used as the characteristic distance in the grain size trend analysis model, which may remove the bias caused by traditional determination of characteristic distance (e.g. based on experience or testing methods). The sediment transport patterns are retrieved on the basis of grain size parameters interpolated with varied distances, which indicates that the small interpolation distance can produce better transport pattern with high resolution and give more detailed information. Hence, the results from geostatistic analysis can be guidable in determination of sampling stations density in the future field work.