Evaluating Transformer and LSTM Frameworks for Prediction in Ungauged Basins
For hydrologists working on prediction in ungauged basins, this work provides a comparative evaluation of architectural inductive biases, showing that LSTM is better suited than encoder-only Transformer for this task.
The study compares encoder-only Transformer and LSTM models for streamflow prediction in ungauged basins, finding that LSTM outperforms Transformer across configurations, and incorporating downstream information improves median NNSE by over 60%.
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. 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. 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. 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