三维非均质农田水力层析扫描研究与应用

The three-dimensional heterogeneity of the variably soil leads to the complexity of the soil water and groundwater movement, which further increases the difficulty to predict solute transport in variably saturated soil. Hydraulic tomography is a recently-developed method that can characterize the three-dimensional distribution of the soil hydraulic properties. Under natural or artificial stimuli, the image of the soil heterogeneity at a particular perspective is revealed based on the observed water or solute information, and the resolution can be enhanced by data fusion. However, the current research of hydraulic tomography mainly focuses on saturated aquifer using pumping as the major approach, which is costly without revealing useful information about the vadose zone. This proposal intends to develop agricultural hydraulic tomography approach for characterizing the detailed three-dimensional heterogeneity of variably saturated soil. Specifically, the natural forces (e.g., precipitation and evapotranspiration) and artificial forces (e.g., irrigation, drainage, pumping, and fertilization) under agricultural conditions are utilized as the stimuli and long-period dynamics of soil water and solute are monitored. The soil survey and geophysical investigation are also employed to enhance the resolution of agricultural hydraulic tomography. The research can provide a scientific tool for accurate prediction of variably saturated flow and solute transport. Based on this, we attempt to make cost-effective water and fertilizer management schemes and prevent groundwater contamination.

变饱和土壤三维空间结构的非均质性造成了土壤水、地下水运动的复杂性，进而导致了溶质运移的复杂性，这使得农田土壤水、地下水及溶质的预测存在较大困难。水力层析是一种刻画含水介质水力参数三维空间分布的方法。在天然或人工刺激下，观测含水介质中的水分或溶质变化，从不同角度对土壤的水力参数空间分布进行扫描，通过数据融合提高非均质性识别分辨率。目前的水力层析多局限于饱和问题，主要采用抽水等手段刺激含水层，成本较高且较难获取非饱和土壤信息。本项目拟发展三维非均质农田水力层析扫描技术，以农田环境下的自然降雨、蒸散发及人工控制的灌溉、排水、抽水、施肥为主要驱动力，观测长时间序列的土壤水、地下水及溶质变化过程，结合土壤采样、地球物理调查等辅助手段，详细描述变饱和土壤水力性质的三维空间结构。成果可为准确预测农田变饱和土壤水和溶质运移提供科学手段，为农业水肥高效管理、防止农业污染提供相关依据。

本项目立足于准确刻画变饱和土壤三维空间结构的非均质性中土壤水的运移，依托广西崇左精准农业研究基地，主要开展了以下研究：（1）探究了初始条件的不确定性在不同的土壤质地、气象条件和剖面长度下的变化规律以及对根区含水率数据价值的影响，并提出了基于可获得观测的土壤水模型初始化优选方案，从而最大的减少初始土壤水分剖面不确定性的影响；（2）基于水量平衡与数据同化技术，提出了一种FDR土壤水传感器观测校准算法，有效地校准了未标定的FDR土壤水传感器观测，从而对土壤水分实现更高精度的模拟和预测，提高了根区含水率的数据价值；（3）通过蒙特卡洛模拟多角度探讨了土壤水分观测在不同的气象条件、下边界条件、土壤质地以及观测位置下的数据价值，提出了一套考虑土壤非均质性的根区土壤含水量时空采样方案，量化不同深度和时间含水率数据价值，可用于指导监测方案；（4）通过设计不同情景探究物理和数据驱动模型模拟土壤水的差异性表现，探究两类土壤水模拟模型的适用性；（5）分析利用土壤水时间序列的频域信息，更好地利用多源信息模拟预测土壤水时间序列。上述研究对指导农业活动、提高农业用水效率、保护农业水环境有重要实际意义。