大量施用化肥提高了烟叶产量，但同时也增加了养分流失和水体污染的风险。深入理解土壤溶质运移规律，对于增加作物产量、提高农业水肥管理水平至关重要。本研究采用空间尺度各异的试验设计和人工模拟降雨试验，以溴离子（Br^-）为示踪剂，系统研究了不同降雨特征（降雨量、降雨强度、降雨滞后时间）和土壤性质（土壤粘粒和砂粒含量、有机碳含量）对土壤溶质运移的影响，以期为烟叶种植期间化肥乃至农药的合理施用提供科学依据，促进烟草种植业的可持续发展。

本研究以云南省3种典型植烟土壤——大理水稻土、红河红壤和玉溪紫色土为研究对象，在当地的代表性烟田各布设了1条长82.5 m、宽1 m的试验样带，在其中均匀划分24个小区，以Br^-为示踪剂，利用人工模拟降雨试验，使3个降雨特征——降雨量、降雨强度和滞后时间（施溴-降雨时间间隔）在样带上作周期性变化，对应周期分别为12、8和4个小区。在此基础上，利用半方差函数和频谱分析探究各降雨特征及土壤性质对土壤溶质运移的影响，并利用线性回归方法和状态—空间模型对各土壤不同深度的Br^-含量空间分布进行模拟，得到的主要结论如下。

（1）3种植烟土壤均没有观察到明显的优先流路径，Br^-主要以基质流形式淋溶。Br^-含量随土壤深度增加逐渐降低；在同一深度，Br^-含量由高到低依次为：红河红壤>大理水稻土>玉溪紫色土。

（2）各土壤Br^-含量受多个因素影响，其水平分布均呈多尺度空间变异，部分周期性变化与降雨特征一致，反映了后者的潜在影响。功率谱分析结果显示，3种土壤几乎所有深度的Br^-含量分布都存在与降雨量一致的周期性变化，大理水稻土、红河红壤和玉溪紫色土部分深度Br^-含量的周期性变化与降雨强度、滞后时间一致。

（3）不同植烟土壤、不同深度Br^-分布的主要影响因素不同。降雨量是几乎所有土层最主要的影响因素，降雨强度、滞后时间主要影响表层土壤Br^-含量，而下层土壤Br^-分布多受土壤性质的影响。总体上，高降雨量可促进Br^-向下迁移；降雨强度越大，滞后时间越短，越不利于Br^-的淋溶。

（4）多元线性回归假定样本间相互独立，仅部分土层Br^-含量可利用多元线性回归模型进行模拟，但拟合效果普遍较差。状态—空间方程充分利用变量自身与变量之间的空间相关关系，对各植烟土壤不同深度Br^-含量的模拟效果明显优于多元线性回归方法。

A heavy application of chemical fertilizers can increase tobacco yield, but also raise the risk of nutrient loss to water bodies and pollution there. It is therefore of great importance to understand solute transport mechanisms in the soils, thereby to enhance crop yield, as well as to improve the agricultural management of water and fertilizers. Using bromide (Br^-) as a tracer, and applying the spatial scale-dependent treatment distribution in a rainfall simulation experiment, the quantitative effects of different rainfall characteristics (i.e., rainfall amount, rainfall intensity, application time delay) and soil properties (i.e., clay and sand contents, soil organic carbon contents) on solute transport were systematically studied, thereby to provide scientific basis for the appropriate applications of fertilizers and pesticides, and to encourage the sustainable development of tobacco production.

Three typical tobacco-planting soils in the Yunnan Province were selected to receive the simulated rainfall in the current study, which were rice paddy soil in the Dali Bai Autonomous Prefecture, red soil in the Honghe Hani and Yi Autonomous Prefecture and purple soil in the Yuxi City. An experimental transect of 82.5 m long and 1 m wide was established in the representative tobacco field of each soil, and 24 plots were design on the transect. Using Br^- as a tracer, the three rainfall characteristics, i.e., rainfall amount, rainfall intensity and application time delay (i.e., time delay between Br^- application and subsequent rainfall) in the rainfall simulation experiment were arranged in a periodically repetitive pattern at three different scales, i.e., 12, 8 and 4 plots, respectively. Semivariance analysis and spectral analysis were employed to quantify the impacts of different rainfall characteristics and soil properties on Br^- leaching, and linear regression and state-space modelling were applied to simulate Br^- distribution at different soil depths. The main findings are concluded as below.

(1) No apparent preferential flow pathway was observed in any of the three tobacco-planting soils investigated. Bromide leaching was dominated by matrix flow at the scale of measurement. No matter for which soil, Br^- concentration generally decreased with soil depth. For the same depth, the Br^- concentration followed the order of red soil > rice paddy soil > purple soil.

(2) The Br^- concentration of each soil was affected by many factors, resulting in a multi-scale variation at each depth. Some periodicities of Br^-, according to the spectral analysis, were consistent with those of the rainfall characteristics, indicating the potential impacts of the latter.  The results of spectral analysis showed that Br^- concentration exhibited consistent oscillations with those of rainfall amount at almost all the depths for each soil. Similar periodicities were also observed in the Br^- concentration of the rice paddy soil, red soil and purple soil at some depths with rainfall intensity and application time delay.

(3) The main influencing factors of Br^- distribution varied among different soils and depths. Rainfall amount was the dominant factor at almost all the depths of each soil, while rainfall intensity and application time delay mainly affected the Br^- distributions near soil surface The Br^- distributions in the lower layers were primarily controlled by soil properties. In general, an elevated amount of rainfall was capable of promoting Br^- leaching; whereas, increasing rainfall intensity and application time delay overall impeded the downward movement of Br^-.

(4) The multiple linear regression (MLR) modeling assumes all the measurements were independent. Significant MLR models were obtained for Br- distribution at some of the depths investigated, no matter for which soil. The performance of the corresponding models were rather poor. In contrast,  state-space modelling took advantage of the spatial dependence inherent in Br^- concentration and its spatial correlations with the influencing factors, thus outperforming MLR in simulating Br^- concentration at each depth for each tobacco-planting soil.