Evaluating Transformer and LSTM Frameworks for Prediction in Ungauged Basins

评估用于无测站流域预测的 Transformer 与 LSTM 框架

Abstract: Watershed networks exhibit convergent topologies in which multiple tributaries merge into downstream channels, integrating diverse upstream hydrological processes. In ungauged basins, the absence of direct observations increases uncertainty and limits the ability to anticipate extreme events.

摘要： 流域网络呈现出汇聚型拓扑结构，即多条支流汇入下游河道，整合了上游多样的水文过程。在无测站流域中，由于缺乏直接观测数据，不确定性增加，限制了对极端事件的预测能力。

This study evaluates whether an encoder-only Transformer provides an advantage over an LSTM for upstream streamflow inference under limited hydrologic information, using retrospective simulations from the NOAA National Water Model (NWM). Across both upstream-only and combined configurations, the LSTM showed stronger overall performance than the Transformer model across the two configurations.

本研究利用美国国家海洋和大气管理局（NOAA）国家水文模型（NWM）的回溯模拟数据，评估了在水文信息有限的情况下，仅含编码器的 Transformer 在上游径流推断方面是否比 LSTM 具有优势。在仅考虑上游和综合配置两种情况下，LSTM 的整体表现均优于 Transformer 模型。

Incorporating downstream information further boosted performance for all models, increasing median NNSE by more than 60%. Rather than treating this as a leaderboard-style comparison, we interpret the experiments as a test of architectural inductive bias for hydrologic sequence inference.

引入下游信息进一步提升了所有模型的性能，使中位数 NNSE（纳什效率系数）提高了 60% 以上。我们并未将此视为排行榜式的竞争，而是将这些实验解读为对水文序列推断架构归纳偏置的测试。

The results indicate that recurrent memory remains better aligned with this upstream reconstruction task than an encoder-only Transformer, while downstream hydrologic context provides a strong auxiliary constraint that substantially improves prediction skill across architectures.

研究结果表明，循环记忆机制在处理此类上游重构任务时，依然比仅含编码器的 Transformer 更具优势；同时，下游水文背景提供了一种强有力的辅助约束，能够显著提升各类架构的预测能力。