EXP-HYDRO: Simulated Daily Runoff from Physics-Informed Deep Learning Models

Five models, including EXP-HYDRO, simulated daily runoff in millimeters per day for three subbasins in the source region of the Yellow River. The data archive includes source code and simulated outputs from a 2023 study by researchers from Tsinghua University. The associated paper was published in Water Resources Research.

Use Cases

• Compare runoff simulation performance between physics-informed and standard deep learning models based on the five provided model outputs.
• Validate hydrological models for alpine catchments based on simulated daily runoff data from the Yellow River source region.
• Benchmark new hybrid deep learning architectures for hydrology based on the provided hybrid-J and hybrid-Z model implementations.

Strengths

• Includes source code for four distinct model architectures (EXP-HYDRO, standard DL, hybrid-J, hybrid-Z), enabling reproducibility.
• Simulated runoff data is provided for three specific geographic subbasins, allowing for spatial analysis.
• Data is directly linked to a peer-reviewed publication in Water Resources Research, providing academic context.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• The simulated data covers only three subbasins, which may limit generalizability to other alpine regions.

Provenance

Source

Tsinghua University

Collection Method

Output from physics-informed deep learning models simulating hydrological processes.

Geography

Three subbasins in the source region of the Yellow River (Alpine catchments).