Diagnosis of observation, background and analysis‐error statistics in observation space

Desroziers, Gérald; Berre, Loïk; Chapnik, Bernard; Poli, Paul

doi:10.1256/qj.05.108

Cited by 704 publications

(1,051 citation statements)

References 17 publications

Supporting

Mentioning

1,031

Contrasting

Order By: Relevance

“…Specifically, observation error incorporates systematic and random errors in instruments and their replacements, errors in data reprocessing and representation error, which arises due to the spatiotemporal incompleteness of observations (Dee and Uppala, 2009;Desroziers et al, 2005). Model error refers mainly to the inadequate representation of physical processes in NWP models (Peña and Toth, 2014;Bengtsson et al, 2007), such as the lack of time-varying surface conditions such as vegetation growth (Zhou and Wang, 2016b;Trigo et al, 2015), and incomplete cloud-precipitation-radiation parameterizations (Fujiwara et al, 2017;Dolinar et al, 2016).…”

Section: Introductionmentioning

confidence: 99%

On the suitability of current atmospheric reanalyses for regional warming studies over China

Zhou

Wang

2018

Atmos. Chem. Phys.

View full text Add to dashboard Cite

Abstract. Reanalyses are widely used because they add value to routine observations by generating physically or dynamically consistent and spatiotemporally complete atmospheric fields. Existing studies include extensive discussions of the temporal suitability of reanalyses in studies of global change. This study adds to this existing work by investigating the suitability of reanalyses in studies of regional climate change, in which land-atmosphere interactions play a comparatively important role. In this study, surface air temperatures (T a ) from 12 current reanalysis products are investigated; in particular, the spatial patterns of trends in T a are examined using homogenized measurements of T a made at ∼ 2200 meteorological stations in China from 1979 to 2010. The results show that ∼ 80 % of the mean differences in T a between the reanalyses and the in situ observations can be attributed to the differences in elevation between the stations and the model grids. Thus, the T a climatologies display good skill, and these findings rebut previous reports of biases in T a . However, the biases in theT a trends in the reanalyses diverge spatially (standard deviation = 0.15-0.30 • C decade −1 using 1 • × 1 • grid cells). The simulated biases in the trends in T a correlate well with those of precipitation frequency, surface incident solar radiation (R s ) and atmospheric downward longwave radiation (L d ) among the reanalyses (r = −0.83, 0.80 and 0.77; p < 0.1) when the spatial patterns of these variables are considered. The biases in the trends in T a over southern China (on the order of −0.07 • C decade −1 ) are caused by biases in the trends in R s , L d and precipitation frequency on the order of 0.10, −0.08 and −0.06 • C decade −1 , respectively. The biases in the trends in T a over northern China (on the order of −0.12 • C decade −1 ) result jointly from those in L d and precipitation frequency. Therefore, improving the simulation of precipitation frequency and R s helps to maximize the signal component corresponding to regional climate. In addition, the analysis of T a observations helps represent regional warming in ERA-Interim and JRA-55. Incorporating vegetation dynamics in reanalyses and the use of accurate aerosol information, as in the Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2), would lead to improvements in the modelling of regional warming. The use of the ensemble technique adopted in the twentieth-century atmospheric model ensemble ERA-20CM significantly narrows the uncertainties associated with regional warming in reanalyses (standard deviation = 0.15 • C decade −1 ).

show abstract

Section: Introductionmentioning

confidence: 99%

On the suitability of current atmospheric reanalyses for regional warming studies over China

Zhou

Wang

2018

Atmos. Chem. Phys.

View full text Add to dashboard Cite

show abstract

“…The innovation vector, d (Equation (2)), and analysis increment, δw a , provide valuable information for assessing the statistical performance and internal consistency of the assimilation system (Desroziers et al, 2005). In this section, we examine mean statistics of d and the analysis residual, r = d − Hδw a , where these vectors, with the time index omitted, are understood to contain the innovation vectors and analysis residuals from all cycles in the 1994-2000 period.…”

Section: Assimilation Statisticsmentioning

confidence: 99%

“…Desroziers et al (2005) discuss how the innovations and analysis increments generated by a data assimilation system can be used to diagnose a posteriori the covariances of observation error and background error in observation space. Assuming that the background and observation errors are mutually uncorrelated, and that their covariance matrices are good approximations to the true error covariance matrices, then the covariance matrix of the innovation vector satisfies…”

Section: Specified Versus Diagnosed σ B and σ Omentioning

confidence: 99%

“…The apparent overestimation of σ o , on the other hand, points to limitations in our simple model of the observation-error covariances, which ignores spatial and temporal correlations and employs flow-independent variance estimates derived from a method that is itself subject to assumptions of questionable validity. Although Equation (29) is used purely for diagnostic purposes in this study, it provides the basis of an iterative algorithm for calibrating σ o using the innovations and analysis increments generated by the assimilation system (Desroziers et al, 2005). In a similar way, Equation (28) can be used to calibrate observationspace values of σ b .…”

Section: Specified Versus Diagnosed σ B and σ Omentioning

confidence: 99%

See 1 more Smart Citation

Ensemble estimation of background‐error variances in a three‐dimensional variational data assimilation system for the global ocean

Daget

Weaver

Balmaseda

2009

Quart J Royal Meteoro Soc

View full text Add to dashboard Cite

This paper studies the sensitivity of global ocean analyses to two flow-dependent formulations of the background-error standard deviations (σ b ) for temperature and salinity in a three-dimensional variational data assimilation (3D-Var) system. The first formulation is based on an empirical parameterization of σ b in terms of the vertical gradients of the background temperature and salinity fields, while the second formulation involves a more sophisticated approach that derives σ b from the spread of an ensemble of background states. The ensembles are created by explicitly perturbing both the surface fluxes (wind stress, fresh water and heat) used to force the model and the observations (temperature and salinity profiles) used in the assimilation process. The two formulations are compared in two cycled 3D-Var experiments for the period 1993-2000. In both experiments, the observation-error standard deviations (σ o ) are geographically dependent and estimated from a model-data comparison prior to assimilation. An additional 3D-Var experiment that employs the parametrized σ b but a simpler σ o formulation, and a control experiment involving no data assimilation, were also conducted and used for comparison.All 3D-Var experiments produce a significant reduction in the mean and standard deviation of the temperature and salinity innovations compared to those of the control experiment. The largest differences between the two σ b formulations occur in the upper 150 m, where the parametrized σ b are notably larger than the ensemble σ b . In this region, the innovation statistics are slightly better for the parametrized σ b . Statistical consistency checks indicate that both schemes underestimate σ b , the underestimation being stronger with the ensemble formulation. The error growth between cycles, however, is much reduced with the ensemble σ b , suggesting that the analyses produced with the ensemble σ b are in better balance than those produced with the parametrized σ b . This claim is supported by independent data comparisons involving model fields not directly constrained by the assimilated temperature and salinity profiles. In particular, sea-surface height (SSH) anomalies in the northwest Atlantic and zonal velocities in the equatorial Pacific are clearly better with the ensemble σ b than with the parametrized σ b . Results also show that while some aspects of those variables are improved with data assimilation (SSH anomalies and currents in the central and eastern Pacific), other aspects are degraded (SSH anomalies in the northwest Atlantic, currents in the western Pacific). Areas for improving the ensemble method and for making better use of the ensemble information are discussed.

show abstract

“…Desroziers et al (2005) proposed a method to diagnose the observation error correlations. Previous studies applied the Desroziers diagnostics and estimated the observation error correlations such as the inter-channel correlations of the satellite radiances (Garand et al 2007;Stewart et al 2013), vertical correlations of radiosondes (Lönnberg and Hollingsworth 1986), and horizontal correlations of the atmospheric motion vectors, sea-surface winds by satellite scatterometers, and radar data (Keeler and Ellis 2000;Bormann et al 2003).…”

Section: Introductionmentioning

confidence: 99%

Data Assimilation with Error-Correlated and Non-Orthogonal Observations: Experiments with the Lorenz-96 Model

Terasaki

Miyoshi

2014

SOLA

View full text Add to dashboard Cite

This study aims to investigate the impact of observation error correlations and non-orthogonal observation operators on analysis accuracy using a chaotic dynamical model known as the Lorenz-96 40-variable model, extending the previous study by Miyoshi et al. using a simple two-dimensional conceptual model. The results corroborate Miyoshi et al.'s conceptual study and show that the analysis is more accurate when the row vectors of a linear observation operator are correlated positively (negatively) with negatively (positively) correlated observation error. The online estimation of the observation error covariance matrix based on the Desroziers diagnostics is successful when we have reasonable a priori knowledge about the observation error correlations.(Citation: Terasaki, K., and T. Miyoshi, 2014: Data assimilation with error-correlated and non-orthogonal observations: Experiments with the Lorenz-96 model. SOLA, 10, 210−213,

show abstract

Diagnosis of observation, background and analysis‐error statistics in observation space

Cited by 704 publications

References 17 publications

On the suitability of current atmospheric reanalyses for regional warming studies over China

On the suitability of current atmospheric reanalyses for regional warming studies over China

Ensemble estimation of background‐error variances in a three‐dimensional variational data assimilation system for the global ocean

Data Assimilation with Error-Correlated and Non-Orthogonal Observations: Experiments with the Lorenz-96 Model

Contact Info

Product

Resources

About