Operational wind yield

Documented changes and trials

All four resources are required for onboarding. Wind heights, grids and source end dates differ. Select a reference in Fit & month selection.

The model-based forecast starts at the first month shown. Hover or focus on a bar for the date and estimated generation on the displayed source basis.

The selected resource controls long-term calibration. Hourly loss modelling and the site power curve use ERA5. The same included months set calibration and historical loss defaults.

Each point shows its observed hours. Incomplete months start excluded; you may include them, but their point and fit use observed hours only. Click a point to include or exclude that month and recalculate the fit and yield. Review hourly data and add comments in Month detail.

Each series retains its source or provider wind height and delivery cadence, shown in the legend. BrightHub histories use ERA5 and CERRA at 100 m and MERRA-2 at 50 m; direct-source histories retain their original definitions. Differences in level are not a ranking of resource quality.

Rolling values average the current month and previous 11 monthly means, with equal weight per month. They require 12 consecutive complete months and are plotted at the final month.

Each bar stacks observed export with the two separate loss estimates. An asterisk marks an excluded or incomplete month. Incomplete bars cover observed hours only. Hover or focus for coverage and exclusion reasons; select a bar for hourly detail.

Select a month to inspect its hourly data. The default selects the latest 36-calendar-month window available for this source. Older eligible months remain shown; use their checkboxes to add them, then Recalculate. Incomplete months start unticked; including one uses its observed hours only. Comments are retained in your scenario and Excel export.

NESO negative BOA identifies BM curtailment. An hourly observation more than two bin standard deviations below the BOA-corrected median receives the full shortfall as additional inferred loss, with no assigned cause.