This website uses cookies. Read about how we use cookies.

Precipitation Nowcasting Research

RMSF error statistics comparing the performance of a nowcast blending extrapolation and NWP with the performances of extrapolation and NWP alone and a persistence forecast.

Rapid update, high resolution, very short range precipitation forecasts are generated by blending extrapolated observations with a recent, 4km resolution forecast from the Unified Model.

Nowcasts are used by a wide range of customers as a source of detailed guidance on the location, extent and timing of imminent, high impact weather events. For example, precipitation nowcasts are utilised by the Met Office and the Environment Agency to mitigate the effects of pluvial and fluvial flooding (FFC).
Whilst significant advances in high resolution NWP prediction have been made in recent years (e.g. the implementation of 4 km and variable (1.5 km) configurations of the Unified Model and further advances including the implementation of a RUC, storm scale (1.5 km) configuration of the UM are planned, there remains a requirement for nowcasting. Nowcasting techniques are capable of exploiting current weather observations more fully (skilfully) in the very short range (lead times < 3 hours) than is achievable with current, operational NWP models.
A suite of nowcasting algorithms is run within the post-processing system. The latter was implemented in 2007 to provide a single source of nowcasts and post-processed, short range forecasts from the 4 km and variable (1.5 km) configurations of the UM. The UKPP system incorporates a precipitation nowcasting algorithm known as STEPS. 
STEPS generates control and 30 member ensemble nowcasts on 15 minute and hourly update cycles respectively. These have a range of six hours and spatial and temporal resolutions of 2 km and 15 minutes respectively. Each nowcast scale-selectively blends extrapolated observations with the most recent forecast from the 4 km UM and a time series of synthetically generated precipitation fields (noise) with space-time statistical properties inferred from weather radar. The noise component serves to account for uncertainties in the evolution of the extrapolation and NWP forecast components and also to downscale the NWP forecast.  
Current R&D is focused on the use of similar techniques to generate seamless, high resolution (2km) precipitation forecast ensembles with a range of several days. These will integrate ensemble precipitation nowcasts with small ensemble NWP precipitation forecasts from the variable (1.5 km) resolution UM and the North Atlantic and European configuration of MOGREPS. The aim is to improve pluvial and fluvial flood warnings issued by the FFC.

Key aims:

  •  To provide high resolution (at least 2km, 15 minute), rapid update analyses and very short range forecasts of surface precipitation rate and accumulation for use by customers including the Met Office's Operations Centre, the Environment Agency and FFC, for pluvial and fluvial flood forecasting and warning.
  • To exploit developments in weather radar and satellite data processing to generate improved analyses of surface precipitation rate and accumulation.
  • To develop techniques for improving precipitation nowcasts and the post-processing of UM forecasts of precipitation to provide more skilful nowcasts and forecasts of precipitation for the benefit of internal and external customers.
  • To develop techniques for quantifying uncertainty in nowcasts and UM forecasts of precipitation and improved ways of communicating this uncertainty to customers involved in pluvial and fluvial flood forecasting and warning.

Current projects:

  • Development, evaluation and operational support for STEPS including migration from the 4 km to variable (1.5 km) configuration of the UM and exploitation of ensembles of radar observations.
  • Development of an algorithm to produce seamless, high resolution (at least 2km) ensemble precipitation forecasts with a range of several days for use in pluvial and fluvial flood forecasting and warning.

Last updated: Feb 21, 2017 1:22 PM