Addressing Model Errors in the UK Met Office

Tim Hewson

Chief Forecaster / Forecasting Research
Met Office, Exeter, UK


The aims of operational forecasting and numerical weather prediction research are similar in that both focus on reducing model forecast error. Activity underway in each category at the new Met Office headquarters in Exeter will be discussed in this talk.

In Exeter, two principal forecasters issue general guidance for the UK and NE Atlantic region for lead times up to about 6 days, which provides the basis for customer-specific forecasts produced elsewhere. The so-called deterministic element of the guidance is generally based on a derived multi-model consensus solution with a second stage of post-processing incorporated to adjust for systematic model errors. Practically, consensus solutions are developed using field modification tools on one model forecast after that model has run. The different tools will be demonstrated, with special focus on adjustments made to the more detailed, shorter range forecasts, highlighting a number of model systematic errors and referring to related research.

Objective and subjective verification measures are applied throughout the lead-time range. These consistently show that the forecaster is, on average, adding value in a cost-effective way. They also provide impetus for NWP development by recording errors that are most commonly addressed, as will be seen.