aiqbench analysis
This study finds that for upstream streamflow inference in ungauged basins, an LSTM outperforms an encoder-only Transformer, suggesting recurrent memory is better aligned with hydrologic sequence reconstruction than current Transformer architectures. While incorporating downstream information significantly boosts prediction skill across all models, the results highlight that architectural inductive bias remains a critical factor when direct observations are unavailable. For practitioners building hydrological models, this indicates that switching to Transformers may not yield immediate gains over established recurrent approaches in this specific domain.
Evaluating Transformer and LSTM Frameworks for Prediction in Ungauged Basins
arXiv:2606.02791v1 Announce Type: new 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. 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
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. 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
Artificial Intelligence (cs.AI)
arXiv:2606.02791 [cs.AI]
(or arXiv:2606.02791v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2606.02791 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 253 words
Related coverage
EEG-FuseFormer: A Transformer-Driven Feature Fusion Framework for Seizure Onset Prediction
arXiv · Jun 5, 2026
From Conceptual Hydrologic Models to Conceptually Interpretable Neural Networks: A Snow-Water Mass-Conserving-Perceptron Framework for Discovering Catchment-Scale Precipitation-Storage-Runoff Representations
arXiv · Jul 30, 2026
An Explainable DistilBERT-BiLSTM-Attention Framework for Binary and Multi-Class Hate Speech Detection
arXiv · Sep 25, 2026
Delta-Based Target Reformulation for Short-Term Electricity Load Forecasting Using LSTM and Transformer Models
arXiv · Jun 17, 2026