Optimal Forecast Combinations Under General Loss Functions and Forecast Error Distributions

Author(s):  
Graham Elliott ◽  
Allan G. Timmermann
2021 ◽  
Author(s):  
Juan Ruiz ◽  
Guo-Yuan Lien ◽  
Keiichi Kondo ◽  
Shigenori Otsuka ◽  
Takemasa Miyoshi

Abstract. Non-Gaussian forecast error is a challenge for ensemble-based data assimilation (DA), particularly for more nonlinear convective dynamics. In this study, we investigate the degree of non-Gaussianity of forecast error distributions at 1-km resolution using a 1000-member ensemble Kalman filter, and how it is affected by the DA update frequency and observation number. Regional numerical weather prediction experiments are performed with the SCALE (Scalable Computing for Advanced Library and Environment) model and the LETKF (Local Ensemble Transform Kalman Filter) assimilating every-30-second phased array radar observations. The results show that non-Gaussianity develops rapidly within convective clouds and is sensitive to the DA frequency and the number of assimilated observations. The non-Gaussianity is reduced by up to 40 % when the assimilation window is shortened from 5 minutes to 30 seconds, particularly for vertical velocity and radar reflectivity.


Wind Energy ◽  
2019 ◽  
Vol 23 (1) ◽  
pp. 45-62
Author(s):  
Jari Miettinen ◽  
Hannele Holttinen ◽  
Bri‐Mathias Hodge

2009 ◽  
Vol 24 (6) ◽  
pp. 1573-1591 ◽  
Author(s):  
Mark DeMaria ◽  
John A. Knaff ◽  
Richard Knabb ◽  
Chris Lauer ◽  
Charles R. Sampson ◽  
...  

Abstract The National Hurricane Center (NHC) Hurricane Probability Program (HPP) was implemented in 1983 to estimate the probability that the center of a tropical cyclone would pass within 60 n mi of a set of specified points out to 72 h. Other than periodic updates of the probability distributions, the HPP remained unchanged through 2005. Beginning in 2006, the HPP products were replaced by those from a new program that estimates probabilities of winds of at least 34, 50, and 64 kt, and incorporates uncertainties in the track, intensity, and wind structure forecasts. This paper describes the new probability model and a verification of the operational forecasts from the 2006–07 seasons. The new probabilities extend to 120 h for all tropical cyclones in the Atlantic and eastern, central, and western North Pacific to 100°E. Because of the interdependence of the track, intensity, and structure forecasts, a Monte Carlo method is used to generate 1000 realizations by randomly sampling from the operational forecast center track and intensity forecast error distributions from the past 5 yr. The extents of the 34-, 50-, and 64-kt winds for the realizations are obtained from a simple wind radii model and its underlying error distributions. Verification results show that the new probability model is relatively unbiased and skillful as measured by the Brier skill score, where the skill baseline is the deterministic forecast from the operational centers converted to a binary probabilistic forecast. The model probabilities are also well calibrated and have high confidence based on reliability diagrams.


Sign in / Sign up

Export Citation Format

Share Document