Model Outputs and Results: Streamflow Predictions for NEON Sites

Predictions and diagnostics from the study 'Virtual gauges: the surprising potential to reconstruct continuous streamflow from strategic measurements' by Vlah et al. 2023. The data includes composite discharge time series, model scores, and predictions for NEON (National Ecological Observatory Network) sites. The archive contains outputs from linear regression and LSTM models, including prediction intervals.

Use Cases

• Benchmarking LSTM and linear regression models for streamflow prediction based on the described model outputs.
• Analyzing uncertainty in discharge estimates based on the provided 95% prediction intervals.
• Studying data gaps in hydrological records based on the included gap statistics.
• Comparing model performance across different NEON sites based on the results.csv files.

Strengths

• Includes prediction intervals (2.5% and 97.5%) for discharge estimates, providing uncertainty quantification.
• Contains outputs from multiple model types, including linear regression and LSTM ensembles, allowing for comparison.
• Data is associated with a specific, cited 2023 research paper, providing academic context.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent for some files; field semantics must be inferred after download.

Provenance

Source

Michael Vlah, associated with the GitHub repository vlahm/neon_q_sim.

Collection Method

Model outputs generated from linear regression and LSTM models applied to NEON and other gauge data.

Time Range

Predictions are available for 2015-2022 where data exists.

Geography

Data pertains to NEON (National Ecological Observatory Network) sites, which are located across the United States.