How to apply bias correction?

In wind resource assessment and energy yield modelling, accurately representing the wind climate at your project site is critical. Whiffle Wind uses high-resolution Large Eddy Simulation to generate site-specific time series and wind resource grids that capture atmospheric dynamics such as turbulence, wake effects, and terrain-induced flow features. These detailed simulations help reduce uncertainty in wind resource estimates and provide a strong foundation for feasibility studies and energy yield predictions.

Like any numerical model, simulation output can show systematic differences when compared with on-site measurements. Minor biases in mean wind speed can become significant when translated into long-term financial projections.

Bias correction is the process of identifying and adjusting these systematic differences to better align simulated data with measured observations. When applied carefully, it improves confidence in model results while preserving the physical consistency of the simulation. This article explains how to set up and run bias correction in Whiffle Wind.

Good to know

Bias correction in Whiffle Wind is currently in beta and available to a select group of users. If you would like to try this feature, please contact [email protected] to request access.

The beta functionality is limited to long-term corrected measurements and long-term statistics outputs from Whiffle Wind.

At this stage, bias correction supports a single measurement location. Support for multiple measurement locations will be added in a future update.

Step 1. Uploading Measurements

1. Navigate to the Measurements tab and click the “+ New measurement” button.

2. Your measurements should be a time series with 10-minute intervals, including both mean wind speed and mean wind direction at each height. The data should meet the following requirements:

   • Long-term corrected wind speeds

   • Timestamps in UTC following the ISO 8601 format

   • NaN values for missing or erroneous data points

3. A template can be downloaded directly from the upload form, or you can prepare your data using the example table below:

Timestamp	MeanSpd_50m	MeanDir_50m	MeanSpd_80m	MeanDir_80m
2015-12-22 00:10:00	4.972	12.8262	8.652	16.2395
2015-12-22 00:20:00	4.972	NaN	8.652	NaN
2015-12-22 00:30:00	NaN	12.8262	8.652	16.2395

Notes:

• The Timestamp column is required.

• Wind speed values should be in meters per second (m/s).

• Wind direction values should be in degrees (°).

Step 2. Filling in the Location and Heights

After uploading your measurements, a form will appear where you can set the name and location of the measurement.

For each column in your CSV file:

• A corresponding row will be created in the form.

• Select the type of measurement: Mean Wind Speed or Mean Wind Direction.

• Enter the height for that measurement.

Requirements:

• Exactly one wind speed and one wind direction must be provided at each height

Finally, mark the measurement as final. A finalised measurement can be used in your simulation but can no longer be edited, ensuring full traceability of your data.

Step 3. Setting Up a Simulation

1. Navigate to the Simulations tab and click “+ New simulation”. For detailed instructions on setting up a simulation, see our guide here.

2. In this beta version, you must place a virtual met mast at the same locations as your measurement points. In future updates, this will be handled automatically when selecting your measurements.

3. Fill in desired WRG output heights that correspond to measurement heights (bias is only corrected at matching heights between the WRG and measurements)

4. On the Overview step, select your previously uploaded measurements to apply bias correction. The simulation will generate:

   • A bias-corrected WRG (Wind Resource Grid)

   • A non-bias-corrected WRG

This allows you to compare results and see the impact of bias correction on your simulation outputs.

FAQ

Does bias correction cost extra?

No. The cost for running a bias correction is the same as a standard statistics run in Whiffle Wind.

Which method is used?

We apply an “Adjust-to-Mast” bias correction that calibrates the modelled wind resource grid to match the measured mast wind climate. For each wind-direction sector, we compute the ratio between the measured and modelled sector mean wind speed (at each mast), then apply a distance-squared (1/d²) weighted blend of those ratios to every location on the site. This results in adjusted wind speeds matching the statistics at the mast locations. The method preserves the model’s spatial patterns while removing systematic bias relative to the measurements.

Why is it not available to everyone?

At Whiffle, we aim to provide a smooth and reliable user experience. The bias correction feature is initially available to a select group of users so we can offer guidance and support during setup. Once the workflow is fully tested and intuitive, the feature will be released to all users.

What if I’m interested in a bias-corrected time series?

You can request a bias-corrected time series by setting up a simulation in Whiffle Wind and selecting Run via support on the Overview step. In the information field, indicate that you would like a bias-corrected time series. Alternatively, you can contact [email protected] to discuss options.