Istvan Szunyogh;
Fan Han
Texas A&M University
The 8th EnKF Data Assimilation Workshop
How Well Can the NCEP Global Ensemble Forecast System Capture the Uncertainty in the Analysis and Forecast of Winter Storm Precipitation?
Talk:
Szunyogh_EnKF8.pdf
We present a new, morphing based technique for the verification of ensemble forecasts of the precipitation. The technique is applied to NCEP operational global forecasts of winter storms in the United States from the 2014/2015 and 2015/2016 winter storm season. The results show that the ensemble forecasts predict the uncertainty in the location of the precipitation systems of the storms with high accuracy, except for a slowly developing systematic error that leads to an unrealistically fast eastward propagation of the storms in the week-two forecasts. This finding suggests that the NCEP ensemble efficiently samples the initial condition and model uncertainties that control the uncertainties in the location of the storms. In contrast, the same forecasts grossly underestimate the uncertainty in the amount of the precipitation at the shorter than 5-day forecast times, which indicates that the ensemble fails to capture the initial condition uncertainties that drive the uncertainties in the prediction of the precipitation amount. We will also discuss the implications of our results for EnKF Data Assimilation.
We present a new, morphing based technique for the verification of ensemble forecasts of the precipitation. The technique is applied to NCEP operational global forecasts of winter storms in the United States from the 2014/2015 and 2015/2016 winter storm season. The results show that the ensemble forecasts predict the uncertainty in the location of the precipitation systems of the storms with high accuracy, except for a slowly developing systematic error that leads to an unrealistically fast eastward propagation of the storms in the week-two forecasts. This finding suggests that the NCEP ensemble efficiently samples the initial condition and model uncertainties that control the uncertainties in the location of the storms. In contrast, the same forecasts grossly underestimate the uncertainty in the amount of the precipitation at the shorter than 5-day forecast times, which indicates that the ensemble fails to capture the initial condition uncertainties that drive the uncertainties in the prediction of the precipitation amount. We will also discuss the implications of our results for EnKF Data Assimilation.
Thanks to McGill's Atmospheric and Oceanic Sciences department for hosting this web-site.