Transforming climate model output to forecasts of wind power production: how much resolution is enough?
Wind power forecasts are useful tools for power load balancing, energy trading and wind farm operations. Long-range monthly-to-seasonal forecasting allows prediction of departures from average weather conditions beyond traditional weather forecast timescales, months in advance. However it has not ye...
| Main Authors: | , , , |
|---|---|
| Format: | Journal article |
| Published: |
Wiley
2017
|
| _version_ | 1797093083534327808 |
|---|---|
| author | Macleod, D Torralba, V Davis, M Doblas-Reyes, F |
| author_facet | Macleod, D Torralba, V Davis, M Doblas-Reyes, F |
| author_sort | Macleod, D |
| collection | OXFORD |
| description | Wind power forecasts are useful tools for power load balancing, energy trading and wind farm operations. Long-range monthly-to-seasonal forecasting allows prediction of departures from average weather conditions beyond traditional weather forecast timescales, months in advance. However it has not yet been demonstrated how these forecasts can be optimally transformed to wind power. The predictable part of a seasonal forecast is for longer monthly averages, not daily averages, but to use monthly averages misses information on variability. To investigate, here we build a model relating average weather conditions to average wind power output, based on the relationship between instantaneous wind speed and power production and incorporating fluctuations in air density due to temperature and wind speed variability. Observed monthly average power output from UK stations is used to validate the model and to investigate the optimal temporal resolution for the data used to drive the model. Multiple simulations of wind power are performed based on reanalysis data, making separate simulations based on monthly, daily and sub-daily averages, using a distribution defined by the mean across the period to incorporate information on variability. Basing the simulation on monthly averages alone is sub-optimal: using daily average winds gives the highest correlation against observations. No improvement over this is gained by using sub-daily averages or including temperature variability. This signifies that to transform seasonal forecasts to wind power a compromise must be made between using the daily averages with debatable skill and the more predictable monthly averages, losing information on day-to-day variability. |
| first_indexed | 2024-03-07T03:55:14Z |
| format | Journal article |
| id | oxford-uuid:c2a56df0-18ab-46c5-80d3-ad447acd3700 |
| institution | University of Oxford |
| last_indexed | 2024-03-07T03:55:14Z |
| publishDate | 2017 |
| publisher | Wiley |
| record_format | dspace |
| spelling | oxford-uuid:c2a56df0-18ab-46c5-80d3-ad447acd37002022-03-27T06:10:31ZTransforming climate model output to forecasts of wind power production: how much resolution is enough?Journal articlehttp://purl.org/coar/resource_type/c_dcae04bcuuid:c2a56df0-18ab-46c5-80d3-ad447acd3700Symplectic Elements at OxfordWiley2017Macleod, DTorralba, VDavis, MDoblas-Reyes, FWind power forecasts are useful tools for power load balancing, energy trading and wind farm operations. Long-range monthly-to-seasonal forecasting allows prediction of departures from average weather conditions beyond traditional weather forecast timescales, months in advance. However it has not yet been demonstrated how these forecasts can be optimally transformed to wind power. The predictable part of a seasonal forecast is for longer monthly averages, not daily averages, but to use monthly averages misses information on variability. To investigate, here we build a model relating average weather conditions to average wind power output, based on the relationship between instantaneous wind speed and power production and incorporating fluctuations in air density due to temperature and wind speed variability. Observed monthly average power output from UK stations is used to validate the model and to investigate the optimal temporal resolution for the data used to drive the model. Multiple simulations of wind power are performed based on reanalysis data, making separate simulations based on monthly, daily and sub-daily averages, using a distribution defined by the mean across the period to incorporate information on variability. Basing the simulation on monthly averages alone is sub-optimal: using daily average winds gives the highest correlation against observations. No improvement over this is gained by using sub-daily averages or including temperature variability. This signifies that to transform seasonal forecasts to wind power a compromise must be made between using the daily averages with debatable skill and the more predictable monthly averages, losing information on day-to-day variability. |
| spellingShingle | Macleod, D Torralba, V Davis, M Doblas-Reyes, F Transforming climate model output to forecasts of wind power production: how much resolution is enough? |
| title | Transforming climate model output to forecasts of wind power production: how much resolution is enough? |
| title_full | Transforming climate model output to forecasts of wind power production: how much resolution is enough? |
| title_fullStr | Transforming climate model output to forecasts of wind power production: how much resolution is enough? |
| title_full_unstemmed | Transforming climate model output to forecasts of wind power production: how much resolution is enough? |
| title_short | Transforming climate model output to forecasts of wind power production: how much resolution is enough? |
| title_sort | transforming climate model output to forecasts of wind power production how much resolution is enough |
| work_keys_str_mv | AT macleodd transformingclimatemodeloutputtoforecastsofwindpowerproductionhowmuchresolutionisenough AT torralbav transformingclimatemodeloutputtoforecastsofwindpowerproductionhowmuchresolutionisenough AT davism transformingclimatemodeloutputtoforecastsofwindpowerproductionhowmuchresolutionisenough AT doblasreyesf transformingclimatemodeloutputtoforecastsofwindpowerproductionhowmuchresolutionisenough |