关键词: 土壤基线值, 数据分布检验, Shapiro-Wilk检验, 正态分布, 异常值

Abstract: The determination of soil baseline values is a critical step in the identification and assessment of ecological environmental damage, as its accuracy directly influences the attribution of pollution responsibility and the outcomes of ecological environmental damage assessments. Guided by GB/T 39791.4—2024 Technical Guidelines for Identification and Assessment of Environmental Damage: General Principles and Key Procedures: Part 4: Investigation and Determination of Soil Environmental Baseline, this study systematically investigated the distribution test of soil background data and the determination of baseline values, using measured soil data from a specific area in northwestern Shanghai as a case study. Through statistical analysis of 14 indicators, including pH, organic matter, and heavy metals, across different soil depths（surface, shallow, and deep layers）, and by comprehensively applying methods such as the Shapiro-Wilk test, graphical methods（histograms, Q-Q plots, P-P plots）, and Dixon outlier test, the data distribution types for each indicator were clarified, and corresponding baseline values for the soil ecological environment were calculated. The results indicated that indicators such as As and Zn exhibited non-normal distributions in the surface layer, reflecting characteristics of exogenous pollution input. The non-normal distributions of Cd and Hg in the shallow layer suggested potential vertical migration and enrichment behaviors of pollutants. Most indicators in the deep soil layers showed normal distributions, representing the natural background levels. This study also found that for pH, the median was more appropriate than the upper limit of the standard-recommended reference value as the baseline to avoid misjudgment. The findings provide a scientific basis for the identification of ecological environmental damage in urban construction waste and other contaminated sites, highlighting the importance of site-specific baselines， and offering optimization suggestions for the practical application of GB/T 39791.4—2024.

Key words: soil baseline values, data distribution test, Shapiro-Wilk test, normal distribution, outliers