The UQ metrics are categorized in three parts: temporal uncertainty, spatial uncertainty, and multiple data sources uncertainty.
- For temporal uncertainty, we consider moving-window based metrics instead of point to point metrics. These metrics reflect the informativeness of the given data.
- For spatial uncertainty, we quantify the uncertainty from wind farm turbine layout. The uncertainty among turbines is converted to a function between uncertainty and distance.
- For uncertainty among multiple data sources, we quantify how the uncertainty spreads among multiple data sources.
UQ metric is summarized as follows:
| Type of Metric | Metric | Description | |
|---|---|---|---|
| Temporal Uncertainty | Moving-window | Entropy | Quantify the uncertainty based on information in the given moving window |
| Standard Deviation | Measure the amount of variation among the given moving window | ||
| Turbulence Intensity | Measure the intensity of wind velocity fluctuation in the given moving window | ||
| Variability Index | Measure the standardized maximum difference in the given moving window | ||
| Spatial Uncertainty | Spatial | Nugget | Quantify small-scale spatial variations within the fields |
| Sill | Measure the magnitude of variation | ||
| Range | The distance beyong which observations are no longer spatially correlated | ||
| Data Source Uncertainty | Moving-window | Correlation | Linear correlation between multiple data sources and the actual |
| MAPE | The average deviation between multiple data sources and the actual | ||
| Distribution | Spread Index | The maximum difference between multiple data sources and the actual | |
| Predictability Index | Identify multiple data sources with above or below average uncertainty |