This is a Preprint and has not been peer reviewed. This is version 2 of this Preprint.
![Skilful probabilistic predictions of UK floods months ahead using machine learning models trained on multimodel ensemble climate forecasts](/media/repos/1/40aab2d6-4c12-4d57-a93c-f005a57ea48c.jpg)
Downloads
Authors
Abstract
Seasonal streamflow forecasts are an important component of flood risk management. Hybrid forecasting methods that predict seasonal streamflow using machine learning models driven by climate model outputs are currently underexplored, yet have some important advantages over traditional approaches using hydrological models. Here we develop a hybrid subseasonal to seasonal streamflow forecasting system to predict the monthly maximum daily streamflow up to four months ahead. We train a random forest machine learning model on dynamical precipitation and temperature forecasts from a multimodel ensemble of 196 members (eight seasonal climate forecast models) from the Copernicus Climate Change Service (C3S) to produce probabilistic hindcasts for 579 stations across the UK for the period 2004-2016, with up to four months lead time. We show that multi-site ML models trained on pooled catchment data together with static catchment attributes are significantly more skilful compared to single-site ML models trained on data from each catchment individually. Considering all initialization months, 60% of stations show positive skill (CRPSS>0) relative to climatological reference forecasts in the first month after initialization. This falls to 41% in the second month, 38% in the third month and 33% in the fourth month.
DOI
https://doi.org/10.31223/X5X405
Subjects
Climate, Hydrology, Physical Sciences and Mathematics
Keywords
floods, flood prediction, seasonal forecasting, machine learning
Dates
Published: 2024-07-12 01:26
Last Updated: 2024-07-23 07:36
Older Versions
License
CC BY Attribution 4.0 International
Additional Metadata
Data Availability (Reason not available):
The input data and scripts that are needed to reproduce the results of this study will be uploaded to a research data repository under an MIT license upon acceptance for publication. They can be made available upon request.
There are no comments or no comments have been made public for this article.