Global Numerical Weather Prediction System Research Articles

Accurate Global Atmospheric State Analysis Using Objective Error Statistics Including Observation Error Dependence on Water Substance Field

AbstractAtmospheric state analysis is a difficult scientific problem but essential for atmospheric sciences. Data assimilation can generate accurate analyses by integrating information on the atmospheric state using probability density functions (PDFs), where the Gaussian approximation is typically used and PDFs are described by error covariance matrices (ECMs). However, ECMs have been estimated empirically, and dependency of the ECMs on meteorological conditions (flow) is only partially represented. These limit atmospheric state analysis accuracy. This is especially problematic for water substance‐sensitive microwave radiances (WS‐MWRs) because of their strong flow dependence. We objectively estimated ECMs of all data including flow‐dependence of the ECMs of WS‐MWRs. Since the ECM of each data is a component of one ECM representing one joint PDF as a whole, it is theoretically better to objectively estimate ECMs of all data, not just a particular data. For WS‐MWRs, we categorized flow into four using water substance amount and estimating an ECM for each category. Numerical experiments using the new ECMs on an operational global numerical weather prediction system show the followings. The new error standard deviations are generally smaller than those of empirical. Standard deviations and interchannel correlations of observation errors of WS‐MWRs increase with water substance amount. The effects of WS‐MWRs on analysis were approximately doubled. The analysis fields differ systematically such as increase of low‐level clouds over cold oceans. The forecast accuracy improved with 95% statistical significance up to 9%. Both the flow dependence of correlation and variance of WS‐MWRs contributed to the improvement of forecast accuracy.

The Rise of Data-Driven Weather Forecasting: A First Statistical Assessment of Machine Learning–Based Weather Forecasts in an Operational-Like Context

Abstract Data-driven modeling based on machine learning (ML) is showing enormous potential for weather forecasting. Rapid progress has been made with impressive results for some applications. The uptake of ML methods could be a game changer for the incremental progress in traditional numerical weather prediction (NWP) known as the “quiet revolution” of weather forecasting. The computational cost of running a forecast with standard NWP systems greatly hinders the improvements that can be made by increasing model resolution and ensemble sizes. An emerging new generation of ML models, developed using high-quality reanalysis datasets like ERA5 for training, allows forecasts that require much lower computational costs and that are highly competitive in terms of accuracy. Here, we compare for the first time ML-generated forecasts with standard NWP-based forecasts in an operational-like context, initialized from the same initial conditions. Focusing on deterministic forecasts, we apply common forecast verification tools to assess to what extent a data-driven forecast produced with one of the recently developed ML models (PanguWeather) matches the quality and attributes of a forecast from one of the leading global NWP systems (the ECMWF IFS). The results are very promising, with comparable accuracy for both global metrics and extreme events, when verified against both the operational IFS analysis and synoptic observations. Overly smooth forecasts, increasing bias with forecast lead time, and poor performance in predicting tropical cyclone intensity are identified as current drawbacks of ML-based forecasts. A new NWP paradigm is emerging relying on inference from ML models and state-of-the-art analysis and reanalysis datasets for forecast initialization and model training.

Assessment for operational assimilation of horizontal line of sight winds from the European Space Agency's Aeolus at the Met Office

AbstractThe European Space Agency's Aeolus satellite was launched in August 2018 and began delivering horizontal line‐of‐sight (HLOS) wind observations in early September 2018. In early 2019, the Met Office began assessing the suitability of the HLOS winds for operational assimilation into its global numerical weather prediction system. We performed a number of assimilation experiments to assess the impact of HLOS wind observations on our global forecasts. We have found that assimilating HLOS winds changes the zonal winds in the analysis fields predominantly in the Tropics and Southern Hemisphere, with the largest changes being in the upper troposphere and lower stratosphere. This has a positive impact on the accuracy of the global weather forecasts, with improvements in the root‐mean‐square error seen throughout the troposphere. Assimilation of Aeolus HLOS winds improves the standard deviation of the observation minus background (a 6 hr forecast) of almost all other observation types, suggesting that the numerical weather prediction model analysis is improved, which consequently improves the 6 hr forecast. In a set of short‐period observation denial experiments, we found that assimilating Aeolus has an impact similar in magnitude to assimilating surface winds from scatterometers. Assimilating winds from the Rayleigh channel has approximately three times the impact that assimilating HLOS winds from the Mie channel does. Both channels contribute a measureable improvement to the global forecast, and we therefore started operational assimilation of winds from the Mie channel in December 2020 and the Rayleigh channel operationally in May 2022.

Reducing a Tropical Cyclone Weak-Intensity Bias in a Global Numerical Weather Prediction System

Abstract The operational Canadian Global Deterministic Prediction System suffers from a weak-intensity bias for simulated tropical cyclones. The presence of this bias is confirmed in progressively simplified experiments using a hierarchical system development technique. Within a semi-idealized, simplified-physics framework, an unexpected insensitivity to the representation of relevant physical processes leads to investigation of the model’s semi-Lagrangian dynamical core. The root cause of the weak-intensity bias is identified as excessive numerical dissipation caused by substantial off-centering in the two time-level time integration scheme used to solve the governing equations. Any (semi)implicit semi-Lagrangian model that employs such off-centering to enhance numerical stability will be afflicted by a misalignment of the pressure gradient force in strong vortices. Although the associated drag is maximized in the tropical cyclone eyewall, the impact on storm intensity can be mitigated through an intercomparison-constrained adjustment of the model’s temporal discretization. The revised configuration is more sensitive to changes in physical parameterizations and simulated tropical cyclone intensities are improved at each step of increasing experimental complexity. Although some rebalancing of the operational system may be required to adapt to the increased effective resolution, significant reduction of the weak-intensity bias will improve the quality of Canadian guidance for global tropical cyclone forecasting. Significance Statement Global numerical weather prediction systems provide important guidance to forecasters about tropical cyclone development, motion, and intensity. Despite recent improvements in the Canadian operational model’s ability to predict tropical cyclone formation, the system systematically underpredicts the intensity of these storms. In this study, we use a set of increasingly simplified experiments to identify the source of this error, which lies in the numerical time-stepping scheme used to solve the model equations. By decreasing numerical drag on the tropical cyclone circulation, intensity predictions that resemble those of other global modeling systems are achieved. This will improve the quality of Canadian tropical cyclone guidance for forecasters around the world.

Impact of Aeolus horizontal line‐of‐sight wind observations on tropical cyclone forecasting in a global numerical weather prediction system

AbstractThe satellite Aeolus, carrying a Doppler Wind Lidar instrument, was launched in August 2018. It is the first mission of the European Space Agency to provide wind profiles globally with high vertical resolutions. The significant positive impact of assimilating the horizontal line‐of‐sight (HLOS) wind data from Aeolus on numerical weather prediction has been reported in many works. In this study, in terms of tropical cyclone (TC) forecasting, the impact of the assimilated data on the Japan Meteorological Agency's global model was investigated in detail. Especially, the direct effect in each analysis resulting from the increment was studied and validated using a reinitialised analysis experiment. The 23 TCs that occurred in the western North Pacific during the summer of 2020 were focused on in the validation period. The largest improvement of track forecasting was found for cases of the NEAR group in which Aeolus passed within 5 degrees of latitude and longitude from a TC centre. The cases of the EAST or WEST groups in which Aeolus passed east or west of the TC centre also indicated some improvements, suggesting that the prediction accuracy of wind upstream of a TC was important for track forecasting. In the NEAR group cases, the increase in positive bias of the TC central pressure was also observed in the cases wherein Aeolus passed inside the maximum wind radius of the TCs. This is explained by the inability to represent the vorticity field in the upper levels in the analyses. One of the possible reasons for this is the insufficient horizontal resolution of observation. We also include some case studies to assess how Aeolus data improved track forecasting and affected intensity forecasting of TCs, specifically.

Improving Wind Power Generation Forecasts: A Hybrid ANN-Clustering-PSO Approach

This study introduces a novel hybrid forecasting model for wind power generation. It integrates Artificial Neural Networks, data clustering, and Particle Swarm Optimization algorithms. The methodology employs a systematic framework: initial clustering of weather data via the k-means algorithm, followed by Pearson’s analysis to pinpoint pivotal elements in each cluster. Subsequently, a Multi-Layer Perceptron Artificial Neural Network undergoes training with a Particle Swarm Optimization algorithm, enhancing convergence and minimizing prediction discrepancies. An important focus of this study is to streamline wind forecasting. By judiciously utilizing only sixteen observation points near a wind farm plant, in contrast to the complex global numerical weather prediction systems employed by the European Center Medium Weather Forecast, which rely on thousands of data points, this approach not only enhances forecast accuracy but also significantly simplifies the modeling process. Validation is performed using data from the Italian National Meteorological Centre. Comparative assessments against both a persistence model and actual wind farm data from Southern Italy substantiate the superior performance of the proposed hybrid model. Specifically, the clustered Particle Swarm Optimization-Artificial Neural Network-Wind Forecasting Method demonstrates a noteworthy improvement, with a reduction in mean absolute percentage error of up to 59.47% and a decrease in root mean square error of up to 52.27% when compared to the persistence model.

Twin‐analysis verification: A new verification approach to alleviate pitfalls of own‐analysis verification

AbstractIn operational numerical weather prediction (NWP), forecast verification against analysis from the same experiment is part of the standard evaluation practice. This “own‐analysis” verification is beneficial in providing complete and uniform spatial coverage but is known to suffer from overestimated skill scores when applied to short‐range forecasts due to the inevitable positive correlation between the forecast and analysis errors. To alleviate this problem, a new approach termed “twin‐analysis” verification is proposed. In this approach, the forecasts are verified against “twin analyses” that are produced by running an independent cycle using the same NWP system as the one used to produce the forecasts, but initializing the cycle using an independent first guess at the very beginning. In this way the error correlation between the forecasts and analyses is reduced, if not completely removed, while preserving the statistical properties of the analyses, allowing for a clearer interpretation of verification. This note reports the results of comparison between “twin‐analysis” and “own‐analysis” verification scores obtained for the Japan Meteorological Agency's global NWP system. It is reported that the two scores disagree up to 48 hr or even longer depending on the regions and elements, which highlights the reliability limit of “own‐analysis” verification. The proposed method does not completely remove the error correlation issue and also requires high computational costs; a practical use of the proposed approach in an operational environment is also discussed.

An Urban Scheme for the ECMWF Integrated Forecasting System: Global Forecasts and Residential CO2 Emissions

AbstractThe impact of urbanization on local weather patterns affects over half the global population. Global numerical weather prediction systems have reached a resolution at which urban conurbations can be spatially resolved, justifying their representation within land surface parameterizations with the aim of improving local predictions. Additionally, real‐time atmospheric monitoring of trace gas emissions can utilize weather variables relevant for urban areas. We investigated whether a simple single‐layer urban canopy scheme can be used within a global forecast model to jointly improve predictions of near‐surface weather variables and residential CO2 emissions. The scheme has been implemented in the Integrated Forecast System used operationally at the European Centre for Medium‐Range Weather Forecasts running at ∼9 km horizontal resolution. First, we selected a suitable urban land cover map (ECOCLIMAP‐SG) based on comparisons with regional data and land surface temperature MODIS retrievals. The urban scheme is verified by providing improved 2 m temperature (∼10%) and 10 m wind (∼17%) RMSE values for both summer and winter months around urban environments. The influence of the scheme was most noticeable at night. Additionally, we have implemented a simple temperature‐dependent residential emissions model to calculate real‐time CO2 heating emissions. These were validated against existing offline products, national reporting and by comparing atmospheric simulations with total column CO2 observations. The results show an improved temporal variability of emissions, which arise from synoptic scale temperature changes. Given the improved predictability from the urban scheme for both weather and emissions, it will be operationally implemented in an upcoming model cycle.

Testing Variational Bias Correction of Satellite Radiance Data in the ACCESS-C: Australian Convective-Scale NWP System

Radiance observations are typically affected by biases that come mainly from instrument error (scanning or calibration) and inaccuracies of the radiative transfer model. These biases need to be removed for successful assimilation, so a bias correction scheme is crucial in the Numerical Weather Prediction (NWP) system. Today, most NWP centres, including the Bureau of Meteorology (hereafter, "the Bureau"), correct the biases through variational bias correction (VarBC) schemes, which were originally developed for global models. However, there are difficulties in estimating the biases in a limited-area model (LAM) domain. As a result, the Bureau's regional NWP system, ACCESS-C (Australian Community Climate and Earth System Simulator-City), uses variational bias coefficients obtained directly from its global NWP system ACCESS-G (Global). This study investigates independent radiance bias correction in the data assimilation system for ACCESS-C. We assessed the impact of using independent bias correction for the LAM compared with the operational bias coefficients derived in ACCESS-G between February and April 2020. The results from our experiment show no significant difference between the control and test, suggesting a neutral impact on the forecast. Our findings point out that the VarBC-LAM strategy should be further explored with different settings of predictors and adaptivity for a more extended period and over additional domains.

Assessment of Retrieved GMI Emissivity over Land, Snow, and Sea Ice in the GEOS System

Abstract Directly assimilating microwave radiances over land, snow, and sea ice remains a significant challenge for data assimilation systems. These data assimilation systems are critical to the success of global numerical weather prediction systems including the Global Earth Observing System–Atmospheric Data Assimilation System (GEOS-ADAS). Extending more surface sensitive microwave channels over land, snow, and ice could provide a needed source of data for numerical weather prediction particularly in the planetary boundary layer (PBL). Unfortunately, the accuracy of emissivity models currently available within the GEOS-ADAS along with other data assimilation systems are insufficient to simulate and assimilate radiances. Recently, Munchak et al. published a 5-yr climatological database for retrieved microwave emissivity from the Global Precipitation Measurement (GPM) Microwave Imager (GMI) aboard the GPM mission. In this work the database is utilized by modifying the GEOS-ADAS to use this emissivity database in place of the default emissivity value available in the Community Radiative Transfer Model (CRTM), which is the fast radiative transfer model used by the GEOS-ADAS. As a first step, the GEOS-ADAS is run in a so-called stand-alone mode to simulate radiances from GMI using the default CRTM emissivity, and replacing the default CRTM emissivity models with values from Munchak et al. The simulated GMI observations using Munchak et al. agree more closely with observations from GMI. These results are presented along with a discussion of the implication for GMI observations within the GEOS-ADAS.

A statistically optimal analysis of systematic differences between Aeolus horizontal line-of-sight winds and NOAA's Global Forecast System

Abstract. The European Space Agency Aeolus mission launched a first-of-its-kind spaceborne Doppler wind lidar in August 2018. To optimize the assimilation of the Aeolus Level-2B (B10) horizontal line-of-sight (HLOS) winds, significant systematic differences between the observations and numerical weather prediction (NWP) background winds should be removed. Total least squares (TLS) regression is used to estimate speed-dependent systematic differences between the Aeolus HLOS winds and the National Oceanic and Atmospheric Administration (NOAA) Finite-Volume Cubed-Sphere Global Forecast System (FV3GFS) 6 h forecast winds. Unlike ordinary least squares regression, TLS regression optimally accounts for random errors in both predictors and predictands. Large, well-defined, speed-dependent systematic differences are found in the lower stratosphere and troposphere in the tropics and Southern Hemisphere. Correction of these systematic differences improves the forecast impact of Aeolus data assimilated into the NOAA global NWP system.

On Two Localized Particle Filter Methods for Lorenz 1963 and 1996 Models

Nonlinear data assimilation methods like particle filters aim to improve the numerical weather prediction (NWP) in non-Gaussian setting. In this manuscript, two recent versions of particle filters, namely the Localized Adaptive Particle Filter (LAPF) and the Localized Mixture Coefficient Particle Filter (LMCPF) are studied in comparison with the Ensemble Kalman Filter when applied to the popular Lorenz 1963 and 1996 models. As these particle filters showed mixed results in the global NWP system at the German meteorological service (DWD), the goal of this work is to show that the LMCPF is able to outperform the LETKF within an experimental design reflecting a standard NWP setup and standard NWP scores. We focus on the root-mean-square-error (RMSE) of truth minus background, respectively, analysis ensemble mean to measure the filter performance. To simulate a standard NWP setup, the methods are studied in the realistic situation where the numerical model is different from the true model or the nature run, respectively. In this study, an improved version of the LMCPF with exact Gaussian mixture particle weights instead of approximate weights is derived and used for the comparison to the Localized Ensemble Transform Kalman Filter (LETKF). The advantages of the LMCPF with exact weights are discovered and the two versions are compared. As in complex NWP systems the individual steps of data assimilation methods are overlaid by a multitude of other processes, the ingredients of the LMCPF are illustrated in a single assimilation step with respect to the three-dimensional Lorenz 1963 model.

A channel selection for the assimilation of CrIS and HIRAS instruments at full spectral resolution

AbstractA new channel selection is proposed for the processing of observations at full spectral resolution (FSR) from the Cross‐track Infrared Sounder (CrIS) and Hyperspectral Infrared Atmospheric Sounder (HIRAS) instruments in the Met Office global numerical weather prediction (NWP) system. The new selection has been derived in order to minimise the error in NWP analysis and has been compared to an existing selection developed at the National Oceanic and Atmospheric Administration (NOAA). Both selections have been tested in the Met Office global NWP system to investigate the use of FSR CrIS compared with the current normal spectral resolution (NSR) set‐up employed in operations. Improvements are obtained for most forecast variables up to 7‐day lead times, with change in root‐mean‐square error (RMSE) ranging from 0.13 to 0.54%, although degradation of tropical temperatures are noted for the NOAA selection. The background fit to observations from independent sounders improves by up to 0.8% with the new selection, outperforming the NOAA selection which results in degradations across most instruments. The assimilation of HIRAS observations does not however benefit the system.

All-sky Data Assimilation of MWTS-2 and MWHS-2 in the Met Office Global NWP System

Microwave radiances from passive polar-orbiting radiometers have been, until recently, assimilated in the Met Office global numerical weather prediction system after the scenes significantly affected by atmospheric scattering are discarded. Recent system upgrades have seen the introduction of a scattering-permitting observation operator and the development of a variable observation error using both liquid and ice water paths as proxies of scattering-induced bias. Applied to the Fengyun 3 Microwave Temperature Sounder 2 (MWTS-2) and the Microwave Humidity Sounder 2 (MWHS-2), this methodology increases the data usage by up to 8% at 183 GHz. It also allows for the investigation into the assimilation of MWHS-2 118 GHz channels, sensitive to temperature and lower tropospheric humidity, but whose large sensitivity to ice cloud have prevented their use thus far. While the impact on the forecast is mostly neutral with small but significant short-range improvements, 0.3% in terms of root mean square error, for southern winds and low-level temperature, balanced by 0.2% degradations of short-range northern and tropical low-level temperature, benefits are observed in the background fit of independent instruments used in the system. The lower tropospheric temperature sounding Infrared Atmospheric Sounding Interferometer (IASI) channels see a reduction of the standard deviation in the background departure of up to 1.2%. The Advanced Microwave Sounding Unit A (AMSU-A) stratospheric sounding channels improve by up to 0.5% and the Microwave Humidity Sounder (MHS) humidity sounding channels improve by up to 0.4%.

The Impact of COVID-19 on Weather Forecasts: A Balanced View.

Aircraft reports are an important source of information for numerical weather prediction (NWP). From March 2020, the COVID‐19 pandemic resulted in a large loss of aircraft data but despite this it is difficult to see any evidence of significant degradation in the forecast skill of global NWP systems. This apparent discrepancy is partly because forecast skill is very variable, showing both day‐to‐day noise and lower frequency dependence on the mean state of the atmosphere. The definitive way to cleanly assess aircraft impact is using a data denial experiment, which shows that the largest impact is in the upper troposphere. The method used by Chen (2020, https://doi.org/10.1029/2020gl088613) to estimate the impact of COVID‐19 is oversimplistic. Chen understates the huge importance of satellite data for modern weather forecasts and raises more alarm than necessary about a drop in forecast accuracy.

The Impact of Length-Scale Variation When Diagnosing the Standard Deviations of Background Error in a 4D-Var System and Filtering Method Investigation

The four-dimensional variational data assimilation (4D-Var) method has been widely employed as an operational scheme in mainstream numerical weather prediction (NWP) centers. In addition to the ensemble data assimilation method, the randomization technique is still used to diagnose the standard deviations of background error in variational data assimilation (VAR) systems; however, such randomization techniques induce sampling noise, which may contaminate the quality of the standard deviations. First, this paper studies the properties of the sampling noise induced by the randomization technique. The results show that the sampling noise is on a small scale displaying high-frequency oscillations around the estimate compared with the estimate and this difference motivates the use of filtering techniques to eliminate the sampling noise effects. The characteristics of the standard deviation field of the control variables are also investigated, and the standard deviation fields of different model parameters have different scales and vary with the vertical model levels. To eliminate such sampling noise, the spectral filtering method used widely in the operational system and a modified spatial averaging approach are investigated. Although both methods have splendid performance in eliminating sampling noise, the spatial averaging approach is more efficient and easier to implement in operational systems. In addition, the optimal filtered results from the spatial averaging approach are dependent on model parameters and vertical levels, which is consistent with the variation in the standard deviation field. Finally, the spatial averaging approach is tested on the operational system at the global scale based on the YH4DVAR and the global NWP system, and the results indicate that the spatial averaging approach has positive effects on both analysis and forecast quality.

Subseasonal Forecasts of Tropical Cyclones in the Southern Hemisphere Using a Dynamical Multimodel Ensemble

AbstractSubseasonal tropical cyclone forecasts from two operational forecast models are verified for the 2017/18 and 2018/19 Southern Hemisphere cyclone seasons. The forecasts are generated using the ECMWF’s Medium- and Extended-Range Ensemble Integrated Forecasting System (IFS), and the Bureau of Meteorology’s seasonal forecasting system ACCESS-S1. Results show the IFS is more skillful than ACCESS-S1, which is attributed to the IFS’s greater ensemble size, increased spatial resolution, and data assimilation schemes. Applying a lagged ensemble with ACCESS-S1 increases forecast reliability, with the optimum number of lagged members being dependent on forecast lead time. To investigate the impacts of atmospheric assimilation at shorter lead times, comparisons were made between the Bureau of Meteorology’s ACCESS-S1 and ACCESS-GE2 systems, the latter a global Numerical Weather Prediction system running with the same resolution and model physics as ACCESS-S1 but using an ensemble Kalman filter for data assimilation. This comparison showed the data assimilation scheme used in the GE2 system gave improvements in forecast skill for days 8–10, despite the smaller ensemble size used in GE2 (24 members per forecast compared to 33). Finally, a multimodel ensemble was created by combining forecasts from both the IFS and ACCESS-S1. Using the multimodel ensemble gave improvements in forecast skill and reliability. This improvement is attributed to complementary spatial errors in both systems occurring across much of the Southern Hemisphere as well as an increase in the ensemble size.

Global Precipitation Forecasts by Merging Extrapolation-Based Nowcast and Numerical Weather Prediction with Locally Optimized Weights

AbstractOver the past few decades, precipitation forecasts by numerical weather prediction (NWP) models have been remarkably improved. Yet, precipitation nowcasting based on spatiotemporal extrapolation tends to provide a better precipitation forecast at shorter lead times with much less computation. Therefore, merging the precipitation forecasts from the NWP and extrapolation systems would be a viable approach to quantitative precipitation forecast (QPF). Although the optimal weights between the NWP and extrapolation systems are usually defined as a global constant, the weights would vary in space, particularly for global QPF. This study proposes a method to find the optimal weights at each location using the local threat score (LTS), a spatially localized version of the threat score. We test the locally optimal weighting with a global NWP system composed of the local ensemble transform Kalman filter and the Nonhydrostatic Icosahedral Atmospheric Model (NICAM-LETKF). For the extrapolation system, the RIKEN’s global precipitation nowcasting system called GSMaP_RNC is used. GSMaP_RNC extrapolates precipitation patterns from the Japan Aerospace Exploration Agency (JAXA)’s Global Satellite Mapping of Precipitation (GSMaP). The benefit of merging in global precipitation forecast lasts longer compared to regional precipitation forecast. The results show that the locally optimal weighting is beneficial.

On the properties of ensemble forecast sensitivity to observations

Evaluating impacts of observations on the skill of numerical weather prediction (NWP) is important. The Ensemble Forecast Sensitivity to Observation (EFSO) provides an efficient approach to diagnosing observation impacts, quantifying how much each observation improves or degrades a subsequent forecast with a given verification reference. This study investigates the sensitivity of EFSO impact estimates to the choice of the verification reference, using a global NWP system consisting of the Non‐hydrostatic Icosahedral Atmospheric Model (NICAM) and the Local Ensemble Transform Kalman Filter (LETKF). The EFSO evaluates observation impacts with the moist total energy norm and with recently proposed observation‐based verification metrics. The results show that each type of observation mainly contributes to the improvement of forecast departures of the observed variable maybe due to the limitation of localization in the EFSO. The EFSO overestimates the fraction of beneficial observations when verified with subsequent analyses, especially for shorter lead times such as 6 h. We may avoid this overestimation to some extent by verifying with observations, analyses from other data assimilation (DA) systems, or analyses of an independent run with the same DA system. In addition, this study demonstrates two important issues possibly leading to overestimating observation impacts. First, observation impacts would be overestimated if we apply relaxation‐to‐prior methods to the initial conditions of the ensemble forecasts in the EFSO; therefore, the ensemble forecasts in the EFSO should be independent of the ensemble forecasts in the DA cycle. Second, deterministic baseline forecasts of the EFSO, which represent the forecast without DA, should be initialized by the ensemble mean of the first guess at the analysis time, not by the previous analysis.

Simulation and diagnosis of observation, model and background error contributions in data assimilation cycling

Data assimilation is usually cycled in time, through a temporal succession of analysis and forecast steps. This implies that forecast errors arise from contributions of observation, model and background errors, which are introduced during successive steps of the cycling. A linearized expansion of forecast errors is here considered, in order to derive expressions and estimates of respective accumulated contributions of these different errors with varying ages. Experiments are conducted in the context of the Météo‐France global numerical weather prediction system ARPEGE.Formal and experimental results indicate that old background errors, which contain all error components arising from corresponding previous cycles, are dampened by successive steps of the cycling, so that they become negligible after a 4‐day period. It is then shown that the variance contribution of recent observation errors tends to converge in time like a power series, within 4 days. This leads the observation error contribution variance to be globally stable in time, which can be interpreted as an approximate compensation between damping of old observation errors and accumulation of recent observation errors. Ensemble experiments and innovation‐based diagnostics are also used to estimate and compare model error contributions with observation error contributions. Such model error diagnostics are likely to be useful e.g. for calibrating model error parametrization schemes in ensemble systems.