Energy

The intelligence you need to anticipate weather-driven

fluctuations in energy supply and demand

Weather is a primary driver for commodity prices in energy, having an impact on both energy consumption (temperature) and production (wind and solar). CFAN’s OmniCast product provides weather and climate forecasts for energy traders, power providers and fund managers. CFAN’s research and product development team has created the most comprehensive and accurate platform available for making decisions related to energy supply and demand​

CFAN’s OmniCast forecast products differentiated from competitor products:

  • Provision of both optimal and probabilistic forecasts, including probabilisitic forecasts of heat/cold waves and wind droughts

  • Proprietary algorithms minimize bias and distributional errors from the global model forecasts, in a calibrated multi-model forecast solution

  • Ensemble clustering increase forecast skill at longer lead times

  • Real-time forecast verification statistics and forecast confidence measures

  • Climate-dynamics based Machine Learning approach to longer range (seasonal) forecasts

 

OmniCast overview

OmniCast provides seamless access to CFAN's complete forecast products for the main energy markets in the U.S. and Europe, on timescales from days to seasons. The OmniCast user interface is based on an interactive layered dashboard approach. Our solutions provide a combination of quick-response, decision-oriented outputs with more complex meteorological graphics.  Easy access is provided to download OmniCast forecast data streams.

 

OmniCast Forecast products:

  • Daily temperature forecasts (1-32 days) (U.S., Europe)

  • Seasonal forecasts of temperature and precipitation (1-6 months)  (U.S.)

  • Renewable energy forecasts 91-32 days)  (Western U.S.)

  • Forecasts of teleconnection regimes  (1-45 days; 1-6 months)

Daily temperature forecasts

 

Forecasts of daily maximum and minimum temperatures are provided twice daily out to 15 days, and twice weekly from days 16-32.  Forecasts are provided for over 900 U.S. cities and over 250 European cities.  Multiple models are used:  ECMWF, GEFS, CFSv2 (subseasonal).

 

For each city, the following temperature forecast information is provided:

  • Table showing daily raw model forecasts, CFAN calibrated forecasts, and CFAN consensus forecast, color coded to show the deviation from the 10 or 30 year climatology for that date

  • Inter-percentile, calibrated multi-model plume view to denote forecast uncertainty

  • Real-time and historical verification statistics for each forecast lead time

  • Heat wave and cold event probabilities

  • Population-weighted heating/cooling degree days

  • Table showing 12 and 24 hour changes in the forecast

 

Insights into forecast uncertainty are provided by plume of the forecast ensemble. Real-time verification statistics are provided for each city, for each daily lead time. Verification statistics are provided for the past 15, 30 and 90 days.

 

City temperature forecast products are illustrated for a summer heatwave that occurred in Portland OR during July 2022. 

Picture1.png
Picture2.png
Picture3.png

Seasonal verification statistics are provided as a function of lead time for each region.  Verification statistics for forecasted U.S. average temperatures for Apr•May•Jun 2022 compare CFAN's forecast with the ECMWF and GEFS ensemble means, GFSX, and climatology.

Picture4.png

Additional dashboard features include:

  • Rapid data feed to provide earliest access to ECMWF forecast data

  • Daily maps of forecasted Tmax and Tmin temperatures

  • Cooling/heating degree day maps

  • Hemispheric maps of upper air temperature and geopotential heights

  • Teleconnection indices: AO, NAO, PNA, WPO, EPO, MJO, GLAAM

  • Quarterly verification statistics for each region

Renewable energy forecasts

 
Wind Turbines
Solar Panel Installation

Matching wind and solar anomaly forecasts with temperature anomalies gives insight as to where and when power demand and supply issues might be most critical.  The need for wind and solar power forecast information is extending to longer projection windows with increasing penetration into the grid and also with diminishing reserve margins to meet peak loads during significant weather events. Maintenance planning and natural gas trading is being influenced increasingly by anticipation of wind and solar generation on timescales of weeks to months.

 

CFAN provides twice weekly forecasts of subseasonal forecasts (7-41 days) for 100 m wind speed and surface solar flux.  CFAN’s renewable energy forecast product for the U.S. includes:

  • Maps of rolling 5 day percentile anomalies

  • Clustered winds based upon weather regime cluster

  • Maps of rolling 5 day wind threshold probability exceedances for selected wind speed intervals.

 

During the period May 20-28, 2019, wind power generation in ERCOT (Texas) averaged 11-14 GW each day, around the clock.  CFAN identified this period of sustained winds beginning on May 6, with a predominance of wind speeds in the range 10-14 m/s. 

4.png
 

Seasonal forecasts

 

CFAN provides global model-based seasonal forecasts (1-6 months) for the U.S. and Europe:

  • Monthly forecasts from global models on the 5th day of the month of monthly temperature and precipitation anomalies

  • Monthly maps of hemispheric height and temperature anomalies; plumes of teleconnections indices; tropical sea surface temperatures; North Atlantic sea surface temperatures and sea level pressure

  • Monthly probabilistic predictions of ENSO (Nino3.4, Nino1.2, Nino3, Nino4) and Modoki; Indian Ocean Dipole and Indian Ocean Basin Mode; Atlantic and Pacific Meridional Mode indices.

3.4.png
IC.png

Seasonal forecasts from the global models show little skill beyond two months unless there is a strong ENSO signal. To address this shortcoming, CFAN has developed a new data-based statistical/dynamical seasonal forecasting system using machine learning methods. Our forecast system is based on climate dynamics research on global circulation patterns in the stratosphere, troposphere and ocean on regional, seasonal weather patterns.

 

CFAN’s statistical/dynamical temperature forecasts project 3-month temperature anomalies at lead times of 3 to 6 months, over 7 U.S. regions and for 1218 local sites.  These forecasts have been produced operationally since October 2017.  The forecasts are produced once per month and disseminated in the form of a report around the 14th of the month.  CFAN’s forecast report issued in May 2019 can be downloaded [link]

Trial subscription

 

CFAN offers trial subscriptions to OmniCast to qualified individuals and organizations. 

To inquire about a trial subscription [link]