NEW CALIBRATED ENSEMBLE RAINFALL PROBABILITY GUIDANCE AND FORECAST AUTOMATION — Australian Meteorological and Oceanographic Society

NEW CALIBRATED ENSEMBLE RAINFALL PROBABILITY GUIDANCE AND FORECAST AUTOMATION (#2042)

Philip Riley 1 , Anja Schubert 1 , Gary Weymouth 1 , Thomas Gale 1 , Timothy Hume 1 , MichaelMichael Foley 1
  1. Bureau of Meteorology, Melbourne, VICTORIA, Australia

Rainfall forecasts have consistently been found to be the most important general weather forecasts for the public and forecasters alike.  Calibrated, combined and downscaled (i.e. post-processed) guidance from Numerical Weather Prediction (NWP) is found to out-perform raw NWP for medium-range weather forecasting on average.

 

A new guidance post-processing system for 3-hourly ensemble rainfall probability which aligns with recently upgraded daily guidance is presented. The new guidance combines a Poor Man's Ensemble of local and international deterministic guidance with the native European Centre for Medium Range Weather Forecasting global weather forecast ensemble. Inputs are combined in probability space in a process analogous to quantile mapping.  Calibration of the guidance is localised and tuned to provide rainfall probabilities at the scale of rain gauges by using daily Australian Water Availability Project analyses close to rain gauges in combination with daily and 3-hourly rain-gauge observations as the 'truth' data.

 

New daily and 3-hourly guidance is intended to be used as the basis for automated forecasts in 'routine' weather conditions.  Increased automation requires compatible daily and 3-hourly guidance of sufficient quality.  Daily rainfall probability skill has been improved by the equivalent of at least 36 hours lead time and shown quality at least as good as official forecasts for many areas in the few months after its introduction. The new 3-hourly guidance also shows large skill gains. Factors affecting performance and uptake of these improved guidance products will be discussed.

#AMOS2019