Search

The use of ensemble forecasts in operational rainfall and flood forecasting systems is rapidly increasing. In the UK, such systems are operated by the Flood Forecasting Centre (FFC) and Scottish Flood Forecasting Service (SFFS) producing ensemble gridded hydrological forecasts out to 6 days. In order to maximise the practical day-to-day use of these systems, for decision-making and warning, duty hydrometeorologists require a sound understanding of both the meteorological and hydrological ensemble forecast skill. A blended Met Office 24-member ensemble precipitation forecast – a mixture of the STEPS nowcast ensemble and STEPS-blended 2.2 km MOGREPS-UK ensemble and 32 km global MOGREPS-G ensemble – drives the Grid-to-Grid (G2G) distributed hydrological model developed by the Centre for Ecology & Hydrology (CEH). G2G uses 15-minute precipitation accumulations as input, and produces river flows at 15-minute intervals on a 1km grid. Phase 1 of the investigation, completed in 2017, formulated and demonstrated a common rainfall and river flow ensemble verification framework. The results gave an initial appreciation of the relative levels of skill in both ensemble rainfall and river flow forecasting systems. In 2018, the verifications of daily and hourly precipitation accumulations will be extended to use 15-minute accumulations. Verifications for three forecast time-horizons – Day 1, Days 2-3 and Days 4-6 – will also be demonstrated. In Phase 1, the sensitivity of verification measures to observation type was illustrated by comparing scores based on radar-only and raingauge-only analyses. Here, in Phase 2, theoretical principles discussed by Ferro (2017) are explored to determine whether a practical application of the theory is possible to gain a more robust measure of forecast performance in the presence of observation uncertainty. Reference Ferro, C.A.T. 2017. Measuring forecast performance in the presence of observation error. Q. J. R. Meteorol. Soc., 143, 2665-2676.

Operational rainfall and flood forecasting systems across the world are increasingly using ensemble approaches. Such systems are operated by the Flood Forecasting Centre (FFC) and Scottish Flood Forecasting Service (SFFS) across Great Britain producing ensemble gridded hydrological forecasts for the next 5-6 days. In order to maximise the practical day-to-day use of these systems for decision-making and warning, duty hydro-meteorologists require a sound understanding of both the meteorological and hydrological ensemble forecast skill. In this work, a common verification framework is defined and used in order to understand the relative levels of skill in both rainfall and river flow forecasting systems. A blended 24-member ensemble precipitation forecast, produced by the Met Office, is used to drive the operational distributed hydrological model in ensemble mode. The hydrological forecasts provide output every 15 minutes out to 6 days on a 1km grid. The blended rainfall forecast is a mixture of the 2.2 km MOGREPS-UK ensemble up to 36h and the 32 km global MOGREPS-G ensemble at longer lead-times. The forecasts are interpolated on to a common 2 km grid and the hydrological model used is the Grid-to-Grid model (G2G) developed by the Centre for Ecology & Hydrology. To establish an upper bound on skill, assessments over a daily lead-time interval are studied first, and will be the focus here. Spatial and regional variations in forecast skill are compared between the precipitation (e.g. daily accumulations) and the river flow forecasts. Also of interest is the impact of catchment size and how to pool and present the skill metrics in a meaningful way for end-users. For precipitation, the impact of observation type: gridded gauge-only analyses and a radar-derived (gauge calibrated) precipitation product, is compared to quantify the uncertainty that comes from the observations. Of particular interest is understanding how the spread in the precipitation forecast is modulated by the downstream hydrological model. Is it inflated, does it remain comparable, or is it reduced? The work aims to establish the basis for a real-time monitoring tool that can assist hydro-meteorologists in their interpretation of operational ensemble forecasts, and facilitate associated decision making processes.