Machine learning methods to improve spatial predictions of coastal wind speed profiles and low-level jets using single-level ERA5 data
<p>Observations of the wind speed at heights relevant for wind power are sparse, especially offshore, but with emerging aid from advanced statistical methods, it may be possible to derive information regarding wind profiles using surface observations. In this study, two machine learning (ML) m...
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Copernicus Publications
2024-04-01
|
| Series: | Wind Energy Science |
| Online Access: | https://wes.copernicus.org/articles/9/821/2024/wes-9-821-2024.pdf |
| _version_ | 1797218094113882112 |
|---|---|
| author | C. Hallgren J. A. Aird S. Ivanell H. Körnich V. Vakkari V. Vakkari R. J. Barthelmie S. C. Pryor E. Sahlée |
| author_facet | C. Hallgren J. A. Aird S. Ivanell H. Körnich V. Vakkari V. Vakkari R. J. Barthelmie S. C. Pryor E. Sahlée |
| author_sort | C. Hallgren |
| collection | DOAJ |
| description | <p>Observations of the wind speed at heights relevant for wind power are sparse, especially offshore, but with emerging aid from advanced statistical methods, it may be possible to derive information regarding wind profiles using surface observations. In this study, two machine learning (ML) methods are developed for predictions of (1) coastal wind speed profiles and (2) low-level jets (LLJs) at three locations of high relevance to offshore wind energy deployment: the US Northeastern Atlantic Coastal Zone, the North Sea, and the Baltic Sea. The ML models are trained on multiple years of lidar profiles and utilize single-level ERA5 variables as input. The models output spatial predictions of coastal wind speed profiles and LLJ occurrence. A suite of nine ERA5 variables are considered for use in the study due to their physics-based relevance in coastal wind speed profile genesis and the possibility to observe these variables in real-time via measurements. The wind speed at 10 <span class="inline-formula"><math xmlns="http://www.w3.org/1998/Math/MathML" id="M1" display="inline" overflow="scroll" dspmath="mathml"><mrow class="unit"><mi mathvariant="normal">m</mi><mspace width="0.125em" linebreak="nobreak"/><mi mathvariant="normal">a</mi><mo>.</mo><mi mathvariant="normal">s</mi><mo>.</mo><mi mathvariant="normal">l</mi><mo>.</mo></mrow></math><span><svg:svg xmlns:svg="http://www.w3.org/2000/svg" width="36pt" height="10pt" class="svg-formula" dspmath="mathimg" md5hash="c66ba9c58375fb02029941ba18d549da"><svg:image xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="wes-9-821-2024-ie00001.svg" width="36pt" height="10pt" src="wes-9-821-2024-ie00001.png"/></svg:svg></span></span> and the surface sensible heat flux are shown to have the highest importance for both wind speed profile and LLJ predictions. Wind speed profile predictions output by the ML models exhibit similar root mean squared error (RMSE) with respect to observations as is found for ERA5 output. At typical hub heights, the ML models show lower RMSE than ERA5 indicating approximately 5 % RMSE reduction. LLJ identification scores are evaluated using the symmetric extremal dependence index (SEDI). LLJ predictions from the ML models outperform predictions from ERA5, demonstrating markedly higher SEDIs. However, optimization utilizing the SEDI results in a higher number of false alarms when compared to ERA5.</p> |
| first_indexed | 2024-04-24T12:12:17Z |
| format | Article |
| id | doaj.art-3729afe3c2fd48579c417ce8f2aee658 |
| institution | Directory Open Access Journal |
| issn | 2366-7443 2366-7451 |
| language | English |
| last_indexed | 2024-04-24T12:12:17Z |
| publishDate | 2024-04-01 |
| publisher | Copernicus Publications |
| record_format | Article |
| series | Wind Energy Science |
| spelling | doaj.art-3729afe3c2fd48579c417ce8f2aee6582024-04-08T10:24:14ZengCopernicus PublicationsWind Energy Science2366-74432366-74512024-04-01982184010.5194/wes-9-821-2024Machine learning methods to improve spatial predictions of coastal wind speed profiles and low-level jets using single-level ERA5 dataC. Hallgren0J. A. Aird1S. Ivanell2H. Körnich3V. Vakkari4V. Vakkari5R. J. Barthelmie6S. C. Pryor7E. Sahlée8Department of Earth Sciences, Uppsala University, Uppsala, SwedenSibley School of Mechanical and Aerospace Engineering, Cornell University, Ithaca, New York, USADepartment of Earth Sciences, Uppsala University, Uppsala, SwedenSwedish Meteorological and Hydrological Institute, Norrköping, SwedenFinnish Meteorological Institute, Helsinki, FinlandAtmospheric Chemistry Research Group, Chemical Resource Beneficiation, North-West University, Potchefstroom, South AfricaSibley School of Mechanical and Aerospace Engineering, Cornell University, Ithaca, New York, USADepartment of Earth and Atmospheric Sciences, Cornell University, Ithaca, New York, USADepartment of Earth Sciences, Uppsala University, Uppsala, Sweden<p>Observations of the wind speed at heights relevant for wind power are sparse, especially offshore, but with emerging aid from advanced statistical methods, it may be possible to derive information regarding wind profiles using surface observations. In this study, two machine learning (ML) methods are developed for predictions of (1) coastal wind speed profiles and (2) low-level jets (LLJs) at three locations of high relevance to offshore wind energy deployment: the US Northeastern Atlantic Coastal Zone, the North Sea, and the Baltic Sea. The ML models are trained on multiple years of lidar profiles and utilize single-level ERA5 variables as input. The models output spatial predictions of coastal wind speed profiles and LLJ occurrence. A suite of nine ERA5 variables are considered for use in the study due to their physics-based relevance in coastal wind speed profile genesis and the possibility to observe these variables in real-time via measurements. The wind speed at 10 <span class="inline-formula"><math xmlns="http://www.w3.org/1998/Math/MathML" id="M1" display="inline" overflow="scroll" dspmath="mathml"><mrow class="unit"><mi mathvariant="normal">m</mi><mspace width="0.125em" linebreak="nobreak"/><mi mathvariant="normal">a</mi><mo>.</mo><mi mathvariant="normal">s</mi><mo>.</mo><mi mathvariant="normal">l</mi><mo>.</mo></mrow></math><span><svg:svg xmlns:svg="http://www.w3.org/2000/svg" width="36pt" height="10pt" class="svg-formula" dspmath="mathimg" md5hash="c66ba9c58375fb02029941ba18d549da"><svg:image xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="wes-9-821-2024-ie00001.svg" width="36pt" height="10pt" src="wes-9-821-2024-ie00001.png"/></svg:svg></span></span> and the surface sensible heat flux are shown to have the highest importance for both wind speed profile and LLJ predictions. Wind speed profile predictions output by the ML models exhibit similar root mean squared error (RMSE) with respect to observations as is found for ERA5 output. At typical hub heights, the ML models show lower RMSE than ERA5 indicating approximately 5 % RMSE reduction. LLJ identification scores are evaluated using the symmetric extremal dependence index (SEDI). LLJ predictions from the ML models outperform predictions from ERA5, demonstrating markedly higher SEDIs. However, optimization utilizing the SEDI results in a higher number of false alarms when compared to ERA5.</p>https://wes.copernicus.org/articles/9/821/2024/wes-9-821-2024.pdf |
| spellingShingle | C. Hallgren J. A. Aird S. Ivanell H. Körnich V. Vakkari V. Vakkari R. J. Barthelmie S. C. Pryor E. Sahlée Machine learning methods to improve spatial predictions of coastal wind speed profiles and low-level jets using single-level ERA5 data Wind Energy Science |
| title | Machine learning methods to improve spatial predictions of coastal wind speed profiles and low-level jets using single-level ERA5 data |
| title_full | Machine learning methods to improve spatial predictions of coastal wind speed profiles and low-level jets using single-level ERA5 data |
| title_fullStr | Machine learning methods to improve spatial predictions of coastal wind speed profiles and low-level jets using single-level ERA5 data |
| title_full_unstemmed | Machine learning methods to improve spatial predictions of coastal wind speed profiles and low-level jets using single-level ERA5 data |
| title_short | Machine learning methods to improve spatial predictions of coastal wind speed profiles and low-level jets using single-level ERA5 data |
| title_sort | machine learning methods to improve spatial predictions of coastal wind speed profiles and low level jets using single level era5 data |
| url | https://wes.copernicus.org/articles/9/821/2024/wes-9-821-2024.pdf |
| work_keys_str_mv | AT challgren machinelearningmethodstoimprovespatialpredictionsofcoastalwindspeedprofilesandlowleveljetsusingsinglelevelera5data AT jaaird machinelearningmethodstoimprovespatialpredictionsofcoastalwindspeedprofilesandlowleveljetsusingsinglelevelera5data AT sivanell machinelearningmethodstoimprovespatialpredictionsofcoastalwindspeedprofilesandlowleveljetsusingsinglelevelera5data AT hkornich machinelearningmethodstoimprovespatialpredictionsofcoastalwindspeedprofilesandlowleveljetsusingsinglelevelera5data AT vvakkari machinelearningmethodstoimprovespatialpredictionsofcoastalwindspeedprofilesandlowleveljetsusingsinglelevelera5data AT vvakkari machinelearningmethodstoimprovespatialpredictionsofcoastalwindspeedprofilesandlowleveljetsusingsinglelevelera5data AT rjbarthelmie machinelearningmethodstoimprovespatialpredictionsofcoastalwindspeedprofilesandlowleveljetsusingsinglelevelera5data AT scpryor machinelearningmethodstoimprovespatialpredictionsofcoastalwindspeedprofilesandlowleveljetsusingsinglelevelera5data AT esahlee machinelearningmethodstoimprovespatialpredictionsofcoastalwindspeedprofilesandlowleveljetsusingsinglelevelera5data |