泰安土壤湿度变化及其与气温和降水的线性相关和非线性因果分析

Soil Moisture Dynamics in Tai’an: Linear Correlation and Nonlinear Causality Analysis with Temperature and Precipitation

摘要: 基于泰安市各县市2015~2024年气象观测站的气温、降水数据和土壤水分观测数据，结合Mann-Kendall非参数检验法和Sen斜率估计法量化土壤湿度的时空变化趋势及其显著性；通过Pearson相关分析揭示气温、降水与土壤湿度的线性统计关联；利用基于核的格兰杰因果方法(KGC)量化气温与降水对土壤湿度的非线性驱动作用。结果表明：(1) 土壤湿度分异：新泰呈显著上升趋势，而东平、宁阳呈显著下降趋势，趋势持续性可能导致未来水分亏缺风险加剧；5~6月为土壤湿度低值期，此期为区域土壤水分保持的关键风险期；(2) 降水为全域关键驱动因子，与土壤湿度显著正相关(新泰站最强)，KGC检验证实其非线性因果影响；(3) 气温调控具有区域异质性：除新泰外，其余站点气温与土壤湿度显著负相关；泰安与肥城两站点同时受降水和气温非线性因果影响；(4) 区域特殊性：新泰存在气温非线性因果影响但统计关联不显著，提示内部复杂机制；东平–宁阳带缺乏气温独立因果贡献，土壤干旱风险主要由降水亏缺主导，气温起依附性放大作用。本研究为区域水资源管理及干旱风险防控提供依据，后续需融合人类活动深化驱动解析。

Abstract: Based on the temperature, precipitation data and soil moisture observation data from meteorological observation stations across counties of Tai’an City from 2015 to 2024, this study employed the Mann-Kendall non-parametric test combined with Theil-Sen estimator to quantify spatiotemporal trends. Pearson correlation analysis was applied to assess linear statistical associations between temperature/precipitation and soil moisture. The Kernel Granger Causality (KGC) was further introduced to quantify nonlinear causal effects of temperature and precipitation on soil moisture. Results demonstrated that: (1) Soil moisture differentiation: Xintai demonstrates a statistically significant increasing trend, while Dongping and Ningyang show a significant decreasing trend, and the continuity of the trend may lead to an increased risk of water deficit in the future; The period from May to June marks a low soil moisture phase, representing a critical risk period for maintaining regional soil water retention. (2) Precipitation as a key driver across the domain is significantly and positively correlated with soil moisture (strongest in Xintai), and KGC confirms its non-linear causal effect; (3) Temperature regulation exhibits regional heterogeneity: Significant negative correlations exist between temperature and soil moisture at all sites except Xintai. Both Tai’an and Feicheng are subject to nonlinear causal effects from precipitation and temperature. (4) Regional specificity: In Xintai, significant nonlinear causal effects of temperature were detected despite statistically insignificant correlations, suggesting complex internal regulatory mechanisms; Dongping and Ningyang lack independent causal contributions from temperature, where soil drought risk is predominantly driven by precipitation deficits, with temperature acting as an amplifying factor. This study provides a scientific basis for regional water resource management and drought risk mitigation. Future research should advance causal attribution by integrating anthropogenic drivers.