Wind Energy

Wind generation is uniquely sensitive to weather — it depends on the cube of wind speed, meaning small forecast errors translate to large generation variance.

The Problem

Wind farm operators face:

• Low-wind periods — generation drops to zero below the cut-in speed (typically 3–4 m/s)
• Extreme wind shutdowns — turbines shut down above cut-out speed (typically 25 m/s)
• Wake effects and turbulence — site-specific phenomena that NWP models don't capture
• Seasonal variability — wind resource varies significantly by season, and inter-annual variability can exceed 15%

How Cliff Horizon Helps

Tier 1 — Wind Risk Analytics

• Calibrated probability of low-wind and high-wind events at the project site
• Generation risk scores accounting for cut-in, rated, and cut-out thresholds
• Ensemble forecast uncertainty bands for wind speed and direction

Tier 2 — Warranted Wind Analytics

• Performance warranty on wind speed predictions at hub height
• Cash payout if prediction accuracy falls outside the agreed Variance Threshold

Tier 3 — Wind Generation Derivative

Pay $150,000 if average wind speed at hub height falls below 5.5 m/s for any calendar month during the contract year.

Modelling Complexity

Wind derivative pricing requires a separate modelling pipeline from temperature. Wind speed follows a Weibull distribution (not Gaussian), requires Box-Cox transformation, AR(4) autocorrelation structure, and non-linear modelling techniques. The mean-reversion speed for wind is time-varying and must be estimated non-parametrically.

Cliff Horizon's engine handles this complexity in its wind pricing module — separate from but integrated with the temperature and rainfall pipelines.

Target Markets

• India — 46 GW installed wind capacity with significant generation variability
• Australia — large wind farms with inter-annual variability
• Middle East — emerging wind markets (Saudi NEOM, Oman)
• East Africa — Lake Turkana and other high-wind sites with limited historical data