发表机构
Canizaro-Livingston Gulf States Center for Environmental Informatics; Naval Research Laboratory（卡尼扎罗-利文斯顿墨西哥湾沿岸各州环境信息学中心; 海军研究实验室）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本研究提出KG-Chronos-2，将冻结的时间序列基础模型与水利工程知识图谱结合，通过残差解码和图检索，在HEC-RAS水面高程预测中显著降低RMSE，验证了知识增强的有效性。
AI 中文摘要
我们研究了将时间序列基础模型与水利工程知识相结合是否能改进HEC-RAS水面高程（WSE）的代理预测。我们提出了KG-Chronos-2，它结合了冻结的Chronos-2预测器与精确状态残差解码、图条件历史检索和输入对齐校正。我们将该方法与持久性模型、残差LSTM、项目条件循环GeoFNO、水力DCRNN风格模型以及冻结的Chronos-2进行了比较。任务特定拟合使用2008年模拟。评估覆盖了2011年和2002年模拟中71个河段4,675个断面的64个固定24小时窗口，基于共享几何。KG-Chronos-2实现了事件平衡均方根误差0.246970（以原生WSE单位计）。相对于冻结的Chronos-2，其RMSE降低了14.13%；相对于水力DCRNN风格模型降低了29.38%；相对于循环GeoFNO降低了39.54%。其与冻结Chronos-2的事件平衡RMSE差异的95%层次自助法区间为[-0.075177, -0.016317]。KG-Chronos-2在六个已完成系统中也实现了最低的活动窗口和最终超前RMSE。这些结果支持将冻结的时间预测器与项目知识相结合，用于固定基准上的热启动HEC-RAS预测。
英文摘要
We investigate whether coupling a time-series foundation model to hydraulic project knowledge improves surrogate forecasting of HEC-RAS water-surface elevation (WSE). We present KG-Chronos-2, which combines a frozen Chronos-2 predictor with exact-state residual decoding, graph-conditioned historical retrieval, and input-aligned correction. We compare the method with persistence, a residual LSTM, project-conditioned recurrent GeoFNO, a hydraulic DCRNN-style model, and frozen Chronos-2. Task-specific fitting uses the 2008 simulation. Evaluation covers 64 fixed 24-hour windows from the 2011 and 2002 simulations at 4,675 cross sections in 71 reaches on a shared geometry. KG-Chronos-2 achieves event-balanced root-mean-square error 0.246970 in native WSE units. It reduces RMSE by 14.13% relative to frozen Chronos-2, 29.38% relative to the hydraulic DCRNN-style model, and 39.54% relative to recurrent GeoFNO. The 95% hierarchical-bootstrap interval for its event-balanced RMSE difference from frozen Chronos-2 is [-0.075177, -0.016317]. KG-Chronos-2 also achieves the lowest active-window and final-lead RMSE among the six completed systems. These results support coupling a frozen temporal predictor to project knowledge for warm-start HEC-RAS forecasting on the fixed benchmark.
Comments9 pages, 4 figures, 4 tables
