A long‐term assessment of precipitation forecast skill using the Fractions Skill Score

摘要: The Fractions Skill Score (FSS) is a spatial verification metric routinely computed in the operational suite. It enables comparison of forecasts different resolutions against common truth (radar rainfall analyses) such way that high-resolution are not penalized for representativeness errors arise from ‘double penalty’ problem. Officially Met Office model precipitation forecast accuracy monitored using Equitable Threat (ETS) at gauge locations. These scores form part basket measures assessing six surface parameters known as UK index, which forms basis making decisions regarding upgrades (especially over UK). used to monitor impact continuous improvements. This framework and methodology underlying it, less appropriate reasons described above. For particular, new long-term monitoring necessary FSS provides potential framework. This paper an objective critique results date. has been shown ‘convection-permitting’ (4 km) Unified Model (MetUM) better than 12 km MetUM (significant 5% level). scale models have sufficient practical skill typically 10 forecasts, forecasting afternoon convection exceeding 4 mm (6 h)−1. use frequency (percentile) thresholds recommended because implicit bias removal this approach provides, any rain period treated ‘the event interest’. © 2011 British Crown Copyright, Office. Published by John Wiley & Sons Ltd.