dc.contributor.author | Kirkwood, C | |
dc.contributor.author | Economou, T | |
dc.contributor.author | Odbert, H | |
dc.contributor.author | Pugeault, N | |
dc.date.accessioned | 2020-06-15T15:45:56Z | |
dc.date.issued | 2021-02-15 | |
dc.description.abstract | Forecasting the weather is an increasingly data intensive exercise. Numerical Weather Prediction (NWP) models are becoming more complex, with higher resolutions, and there are increasing numbers of different models in operation around the world. While the forecasting skill of NWP models continues to improve, the number and complexity of these models poses a new challenge for the operational meteorologist: how should the information from all available models, each with their own unique biases and limitations, be combined in order to provide stakeholders with well-calibrated probabilistic forecasts to use in decision making? In this paper, we use a road surface temperature forecasting example to demonstrate a three-stage framework that uses machine learning to bridge the gap between sets of separate forecasts from NWP models and the 'ideal' forecast for decision support: probabilities of future weather outcomes. First, we use Quantile Regression Forests to learn the error profile of each numerical model, and use these to apply empirically-derived probability distributions to forecasts. Second, we combine these probabilistic forecasts using quantile averaging. Third, we interpolate between the aggregate quantiles in order to generate a full predictive distribution, which we demonstrate has properties suitable for decision support. Our results suggest that this three stage approach provides an effective and operationally viable framework for the cohesive post-processing of weather forecasts across multiple models and lead times in order to produce well-calibrated probabilistic forecasts. | en_GB |
dc.description.sponsorship | Met Office | en_GB |
dc.description.sponsorship | Engineering and Physical Sciences Research Council (EPSRC) | en_GB |
dc.identifier.citation | Vol. 379 (2194), article 20200099 | en_GB |
dc.identifier.doi | 10.1098/rsta.2020.0099 | |
dc.identifier.grantnumber | 2071900 | en_GB |
dc.identifier.uri | http://hdl.handle.net/10871/121448 | |
dc.language.iso | en | en_GB |
dc.publisher | Royal Society | en_GB |
dc.subject | data integration | en_GB |
dc.subject | uncertainty quantification | en_GB |
dc.subject | quantile regression | en_GB |
dc.subject | model stacking | en_GB |
dc.subject | decision theory | en_GB |
dc.subject | artificial intelligence | en_GB |
dc.title | A framework for probabilistic weather forecast post-processing across models and lead times using machine learning | en_GB |
dc.type | Article | en_GB |
dc.date.available | 2020-06-15T15:45:56Z | |
dc.identifier.issn | 1364-503X | |
dc.description | This is the final version. Available on open access from the Royal Society via the DOI in this record | en_GB |
dc.description | Data Accessibility. Our dataset has been made available with permission from the Met Office
and Highways England, for which we are grateful. It is available for download along with
several other open weather forecast post-processing datasets collated by Haupt et al. [39] at
https://doi.org/10.6075/J08S4NDM. In addition, the code for this study can be accessed at
https://github.com/charliekirkwood/mlpostprocessing. | en_GB |
dc.identifier.journal | Philosophical Transactions A: Mathematical, Physical and Engineering Sciences | en_GB |
dc.rights.uri | https://creativecommons.org/licenses/by/4.0/ | en_GB |
dcterms.dateAccepted | 2020-06-12 | |
exeter.funder | ::Met Office | en_GB |
rioxxterms.version | VoR | en_GB |
rioxxterms.licenseref.startdate | 2020-06-12 | |
rioxxterms.type | Journal Article/Review | en_GB |
refterms.dateFCD | 2020-06-15T09:52:38Z | |
refterms.versionFCD | AM | |
refterms.dateFOA | 2021-02-19T12:00:12Z | |
refterms.panel | B | en_GB |