The present paper summarizes the methodologies used at the European Centre for Medium-Range Weather Forecasts (ECMWF), the Meteorological Service of Canada (MSC), and the National Centers for Environmental Prediction (NCEP) to simulate the effect of initial and model uncertainties in ensemble forecasting. The characteristics of the three systems are compared for a 3-month period between May and July 2002. The main conclusions of the study are the following:the performance of ensemble prediction systems strongly depends on the quality of the data assimilation system used to create the unperturbed (best) initial condition and the numerical model used to generate the forecasts;a successful ensemble prediction system should simulate the effect of both initial and model-related uncertainties on forecast errors; andfor all three global systems, the spread of ensemble forecasts is insufficient to systematically capture reality, suggesting that none of them is able to simulate all sources of forecast uncertainty.The relative strengths and weaknesses of the three systems identified in this study can offer guidelines for the future development of ensemble forecasting techniques.

CERA‐20C is a coupled reanalysis of the twentieth century which aims to reconstruct the past weather and climate of the Earth system including the atmosphere, ocean, land, ocean waves, and sea ice. This reanalysis is based on the CERA coupled atmosphere‐ocean assimilation system developed at ECMWF. CERA‐20C provides a 10 member ensemble of reanalyses to account for errors in the observational record as well as model error. It benefited from the prior experience of the retrospective atmospheric analysis ERA‐20C. The dynamical model and the data assimilation systems initially developed for NWP had been modified to take into account the evolution of the radiative forcing and the observing system. To limit the impact of changes in the observing system throughout the century, only conventional surface observations have been used in the atmosphere. CERA‐20C improves the specification of the background and the observation errors, two key elements to ensure a consistent weighting of the uncertainties across geophysical variables, space, and time. The quality of CERA‐20C has been evaluated against other centennial reanalyses and independent observations. Although CERA‐20C inherits some limitations of ERA‐20C to represent correctly the tropical cyclones in the first part of the century, it shows significant improvements in the troposphere, compared to ERA‐20C and 20CRv2c (the twentieth century reanalysis produced by NOAA/CIRES). A preliminary study of the climate variability in CERA‐20C has been carried out. CERA‐20C improves on the representation of atmosphere‐ocean heat fluxes and mean sea level pressure compared to previous uncoupled ocean and atmospheric historical reanalyses performed at ECMWF.

In this paper, we review the use of satellite-based remote sensing in combination with in situ data to inform Earth surface modelling. This involves verification and optimization methods that can handle both random and systematic errors and result in effective model improvement for both surface monitoring and prediction applications. The reasons for diverse remote sensing data and products include (i) their complementary areal and temporal coverage, (ii) their diverse and covariant information content, and (iii) their ability to complement in situ observations, which are often sparse and only locally representative. To improve our understanding of the complex behavior of the Earth system at the surface and sub-surface, we need large volumes of data from high-resolution modelling and remote sensing, since the Earth surface exhibits a high degree of heterogeneity and discontinuities in space and time. The spatial and temporal variability of the biosphere, hydrosphere, cryosphere and anthroposphere calls for an increased use of Earth observation (EO) data attaining volumes previously considered prohibitive. We review data availability and discuss recent examples where satellite remote sensing is used to infer observable surface quantities directly or indirectly, with particular emphasis on key parameters necessary for weather and climate prediction. Coordinated high-resolution remote-sensing and modelling/assimilation capabilities for the Earth surface are required to support an international application-focused effort.

TIGGE, the THORPEX Interactive Grand Global Ensemble, is a World Weather Research Programme project to accelerate the improvements in the accuracy of 1‐day to 2‐week high‐impact weather forecasts. This report discusses some preliminary results from predictability studies based on the ensemble data exchanged within TIGGE, and available at the TIGGE archive centres. In the first part of this work, the key characteristics of the eight ensemble systems available in the TIGGE database at the time of writing (December 2007) are compared, and the strengths and weaknesses of each system are highlighted. Then, issues related to the generation of multi‐model/multi‐analysis ensemble products are discussed, and some preliminary results on the potential value of combining different ensembles to generate medium‐range products with a grand multi‐model/multi‐analysis global ensemble are presented.One of the key results documented in this work is the large difference between the performance of the single ensembles: for Z500 over the Northern Hemisphere, in the medium‐range (say around forecast day 5), the difference between the worst and the best control or ensemble‐mean forecasts is about 2 days of predictability, while the difference between the worst and the best probabilistic predictions can be larger, about 3 days of predictability. Another key result has been the quantification of the difference between the skill of the best ensemble system and a combined ensemble generated considering up to four different ensemble systems. Results have indicated that the difference is very small in areas where the European Centre (EC) ensemble system has a well tuned ensemble spread, equivalent to less than 6 hours of predictability in the medium range, while it is larger and more detectable in areas where the EC system has a too low ensemble spread (e.g. in the Tropics). Copyright © 2008 Royal Meteorological Society

Ensemble forecasting is a feasible method to integrate a deterministic forecast with an estimate of the probability distribution of atmospheric states. At the European Centre for Medium-Range Weather Forecasts (ECMWF), the Ensemble Prediction System (EPS) comprises 32 perturbed and 1 unperturbed nonlinear integrations, at T63 spectral triangular truncation and with 19 vertical levels. The perturbed initial conditions are generated using the most unstable directions growing over a 48-h time period, computed at T42L19 resolution. This work describes the performance of the ECMWF EPS during the first 21 months of daily operation, from 1 May 1994 to 31 January 1996, focusing on the 500-hPa geopotential height fields. First, the EPS is described, and the validation approach followed throughout this work is discussed. In particular, spread and skill distribution functions are introduced to define a more integral validation methodology for ensemble prediction. Then, the potential forecast skill of ensemb... Abstract Ensemble forecasting is a feasible method to integrate a deterministic forecast with an estimate of the probability distribution of atmospheric states. At the European Centre for Medium-Range Weather Forecasts (ECMWF), the Ensemble Prediction System (EPS) comprises 32 perturbed and 1 unperturbed nonlinear integrations, at T63 spectral triangular truncation and with 19 vertical levels. The perturbed initial conditions are generated using the most unstable directions growing over a 48-h time period, computed at T42L19 resolution. This work describes the performance of the ECMWF EPS during the first 21 months of daily operation, from 1 May 1994 to 31 January 1996, focusing on the 500-hPa geopotential height fields. First, the EPS is described, and the validation approach followed throughout this work is discussed. In particular, spread and skill distribution functions are introduced to define a more integral validation methodology for ensemble prediction. Then, the potential forecast skill of ensemb...

One of the crucial aspects of the design of an ensemble prediction system is the definition of the ensemble of initial states. This work investigates the use of singular vectors, an ensemble of analyses, and a combination of the two types of perturbations in the ECMWF operational ensemble prediction system. First, the similarity between perturbations generated using initial‐time singular vectors (SVs) and analyses from the ensemble data assimilation (EDA) system is assessed. Results show that the EDA perturbations are less localized geographically and have a better coverage of the Tropics. EDA perturbations have also smaller scales than SV‐based perturbations, and have a less evident upshear vertical tilt, which explains why they grow less with forecast time. Then, the use of EDA‐based perturbations in the ECMWF ensemble prediction system is studied. Results indicate that if used alone, EDA‐based perturbations lead to an under‐dispersive and less skilful ensemble then the one based on initial‐time SVs only. Combining the EDA and the initial‐time SVs gives a system with a better agreement between ensemble spread and the error of the ensemble mean, a smaller ensemble‐mean error and more skilful probabilistic forecasts than the current operational system based on initial‐time and evolved SVs. Copyright © 2008 Royal Meteorological Society