Synthetic Forecasts of Renewable Energy Supply and Demand across the CONUS: Initial Release

This dataset contains a set of forecasts for variables associated with renewable energy supply and energy demand across the contiguous United States (CONUS). These include: streamflow into 1111 hydropower facilities wind power generation at 92 buses solar power generation at 276 buses load at 54 balacing authorities Streamflow forecasts are provided daily, while wind, solar, and load forecasts are provided hourly. All forecasts extend out 7 days into the future, with new forecasts issued each day for flow and each hour for wind, solar, and load. Three sets of forecasts are available: Baseline (persistence) forecasts: these forecasts are based on climatology, but account for observed, short-term deviations from climatology using an autoregressive model. Perfect forecasts: these forecasts assume perfect foresight of each variable over the forecast horizon Synthetic forecasts: these forecasts are a weighted average between the baseline and perfect forecasts. They represent a scenario of forecast improvement over the baseline. Two parameters govern the degree of improvement: weight: this parameter determines the percent improvement for the 1-day-ahead forecast from the baseline to the perfect forecast scenario.For example, a value of 0.5 for the weight would indicate that the synthetic forecast at a 1-day lead will be half way between the baseline and perfect forecast values at a 1-day lead. half-life: forecast improvements at shorter lead times are generally greater than forecast improvements at longer lead times. The half-life parameter embeds this behavior in the synthetic forecasts. The value of this parameter indicates the lead time when the forecast improvement for 1-day-ahead forecast is halved. So, for instance, if the weight parameter indicates a 50% improvement in the 1-day-ahead forecast, and the half-life parameter equals 5 days, then the improvement for 5-day-ahead forecasts will be 25% (i.e., 25% of the way from the baseline forecast to the perfect forecast at a 5-day lead). A total of 7 forecast scenarios are created in total, based on different combinations of the weight and half-life parameters. Note that the scenarios are developed independently for each variable, enabling a much larger set of unqiue combinations of forecast skill across the wind, solar, load, and flow variables. Forecast Scenario Decay of Forecast Improvement (half-life) Wind Solar Load Flow 1 NA Perfect Perfect Perfect Perfect 2 NA Baseline Baseline Baseline Baseline 3 2-day 25% 25% 25% 25% 4 2-day 50% 50% 50% 50% 5 2-day 75% 75% 75% 75% 6 5-day 50% 50% 50% 50% 7 7-day 50% 50% 50% 50% Do to space limitations, this dataset has been split across multiple repositories. While the wind and solar data can be found in this repository, the load data can be found here: 10.5281/zenodo.15799553 And the flow data can be found here: 10.5281/zenodo.16845587