TECHNIQUES OF TREND ANALYSIS FOR MONTHLY WATER-QUALITY DATA
TECHNIQUES OF TREND ANALYSIS FOR MONTHLY WATER-QUALITY DATA
复制标题
DOI：
10.1029/wr018i001p00107
复制
发表时间：
1982-01-01
影响因子：
5.4
通讯作者：
SMITH, RA
中科院分区：
文献类型：
--
作者：
HIRSCH, RM;SLACK, JR;SMITH, RA
水质时间序列的一些复杂的特性是不自然的分布，季节性，流量相关性，缺失值，低于检测限的值和序列相关性。提出了技术，适合在面对的并发症列出的每月水质数据的单调趋势的探索性分析。第一个程序，季节性Kendall检验，是适用于具有季节性、缺失值或报告值小于的数据集的趋势非参数检验。在现实的随机过程（表现出季节性，偏态和序列相关性），它是强大的参数替代品相比，虽然季节性肯德尔测试或替代品不能被认为是一个确切的测试中存在的序列相关性。第二个程序，季节性肯德尔斜率估计，是一个估计的趋势幅度。它是线性趋势斜率的无偏估计量，并且在数据高度偏斜的情况下比回归估计量具有更高的精度，但在数据正态的情况下精度略低。第三个程序提供了一种测试成分浓度和流量之间关系随时间变化的方法，避免了识别水质趋势的问题，这些趋势是观察到的特定排放序列的人为因素（例如，干旱影响）。在该方法中，流量调整浓度被定义为基于浓度对流量的某种函数的回归的残差（实际值减去条件期望值）。这些流量调整后的浓度，也可能是季节性和非自然的，可以通过使用季节性肯德尔测试来测试趋势。
Some characteristics that complicate the analysis of water quality time series are unnatural distributions, seasonality, flow relatedness, missing values, values below the limit of detection and serial correlation. Presented were techniques that were suitable in the face of the complications listed for the exploratory analysis of monthly water quality data for monotonic trends. The 1st procedure, the seasonal Kendall test, was a nonparametric test for trend applicable to data sets with seasonality, missing values or values reported as less than. Under realistic stochastic processes (exhibiting seasonality, skewness and serial correlation), it was robust in comparison to parametric alternatives, although the seasonal Kendall test or the alternatives can not be considered an exact test in the presence of serial correlation. The 2nd procedure, the seasonal Kendall slope estimator, was an estimator of trend magnitude. It was an unbiased estimator of the slope of a linear trend and had higher precision than a regression estimator where data were highly skewed but somewhat lower precision where the data were normal. The 3rd procedure provided a means for testing for change over time in the relationship between constituent concentration and flow, avoiding the probem of identifying trends in water quality that were artifacts of the particular sequence of discharges observed (e.g., drought effects). In this method a flow-adjusted concentration was defined as the residual (actual minus conditional expectation) based on a regression of concentration on some function of discharge. These flow-adjusted concentrations, which may also be seasonal and unnatural, could be tested for trend by using the seasonal Kendall test.