Rapid infiltration following precipitation may result in groundwater contamination from surface contaminants or pathogens. In fractured rock, contaminants can migrate rapidly to points of groundwater withdrawals. In contrast to the temporal availability of groundwater quality chemical indicators, meteorological and groundwater level observations are available in real-time to estimate time-varying recharge, which can act as a surrogate to identify periods of rapid infiltration that may indicate contamination susceptibility. Estimating recharge using methods, such as base-flow recession, unsaturated infiltration models, or Water-Table Fluctuations (WTF), cannot capitalize on currently available technologies and telecommunication infrastructure to conduct real-time recharge estimation at scales relevant to characterizing rapid infiltration. We present a linear, physics-based State-Space (SS) model of one-dimensional infiltration to estimate recharge, which includes preferential and diffuse-flow to the water table. The model can take advantage of real-time data for water-table altitude, precipitation, and evapotranspiration. Model parameters are calibrated over an observation period, and the Kalman Filter (KF) is subsequently applied to continuously update the observed (water-table altitude) and unobserved (groundwater recharge) system states and predict future states as new data become available. The SS/KF algorithm is demonstrated at the Masser Groundwater Recharge Site in Pennsylvania, USA and comparisons are made with recharge estimates from WTF methods. Model results indicate that the frequency of observations (daily versus sub-daily) dictates the allocation between preferential and diffuse flow. Additionally, because infiltration processes encompass many nonlinearities, model parameters estimated from observation periods need to be updated at least seasonally to account for changing recharge conditions.

中文翻译：

---

估计和预测断裂岩石中随时间变化的地下水补给：具有优先和扩散流到地下水位的状态空间公式

降水后的快速渗透可能会导致地下水受到地表污染物或病原体的污染。在破裂的岩石中，污染物可以迅速迁移到地下水抽取点。与地下水质量化学指标的时间可用性相反，气象和地下水位观测可实时用于估计随时间变化的补给，这可以作为替代物来识别可能表明污染易感性的快速渗透时期。使用基流衰退、非饱和渗透模型或水位波动 (WTF) 等方法估算补给量无法利用当前可用的技术和电信基础设施在与表征快速渗透相关的尺度上进行实时补给估算。我们提出了一个线性的、基于物理的状态空间 (SS) 一维渗透模型来估计补给，其中包括到地下水位的优先和扩散流。该模型可以利用地下水位高度、降水和蒸发量的实时数据。在观测期间校准模型参数，随后应用卡尔曼滤波器 (KF) 不断更新观测到的（地下水位高度）和未观测到的（地下水补给）系统状态，并在新数据可用时预测未来状态。SS/KF 算法在美国宾夕法尼亚州的 Masser 地下水补给站进行了演示，并与 WTF 方法的补给估计进行了比较。模型结果表明观察频率（每日与次日）决定了优先流和扩散流之间的分配。此外，由于下渗过程包含许多非线性，从观察期估计的模型参数需要至少按季节更新，以考虑不断变化的补给条件。