Abstract:When estimating observation‐error covariance matrices using observation‐minus‐background and observation‐minus‐analysis residuals obtained from data assimilation systems, the error comes from two sources. The first is the indirect sampling error caused by using a finite sample of residual vectors, which depends largely on the ratio of vector size to sample size. The second error is due to misspecification of assumed observation‐ and background‐error statistics in the assimilation system, such as incorrectly assuming a correlated observation‐error covariance matrix is diagonal. Specification of the observation‐error covariance matrix for data assimilation systems affects the observation information content retained by the analysis, particularly for observations known to have correlated observation errors (e.g., geostationary satellite and Doppler radar data). A widely adopted approach for estimating observation‐error covariance matrices uses observation‐minus‐background and observation‐minus‐analysis residuals, which are routinely produced by most data assimilation systems. Although this approach is known to produce biased and noisy estimates, due to sampling and misspecification errors, there has been no systematic study of sampling errors with this approach to date. Furthermore, the eigenspectrum of the estimated observation‐error covariance matrix is known to influence the analysis information content and numerical convergence of variational assimilation schemes. In this work, we provide new theorems for the sampling error and eigenvalues of the estimated observation‐error covariance matrices with this approach. We also conduct numerical experiments to illustrate our theoretical results. We find that this method produces large sampling errors if the true observation‐error standard deviation is large, while the other error characteristics, including the true background‐error standard deviation and observation‐ and background‐error correlation length‐scales, have a relatively small effect. We also find that the smallest eigenvalues of the estimated matrices may be smaller or larger than the true eigenvalues, depending on the assumed and true observation‐ and background‐error statistics. These results may provide insights for practical applications: for example, in deciding on appropriate sample sizes and choosing parameters for matrix reconditioning techniques.

On the optimal background error covariances: Different scale errors' contribution

INFLUENCE OF DIFFERENT-SCALE ERRORS INTERACTIONS ON ANALYSIS AND FORECAST OF REGIONAL NWP MODEL

Spectral Characteristics of Background Error Covariance and Multiscale Data Assimilation

Background Error Covariance with Balance Constraints for Aerosol Species and Applications in Variational Data Assimilation

Cloud-dependent Piecewise Assimilation Based on a Hydrometeor-Included Background Error Covariance and Its Impact on Regional Numerical Weather Prediction

Background Error Covariance Statistics of Hydrometeor Control Variables Based on Gaussian Transform

THE EFFECT OF INITIAL MESOSCALE MESOSCALE FEATURE ERRORS ON THE PREDICTION OF PRECIPITATION

Tuning of Background Error Statistics Through Sensitivity Experiments and Its Impact on Typhoon Forecast

Incorporating a Large‐Scale Constraint into Radar Data Assimilation to Mitigate the Effects of Large‐Scale Bias on the Analysis and Forecast of a Squall Line over the Yangtze‐Huaihe River Basin

Effects of Large-Scale Constraint and Constraint Variables on the High-Frequency Assimilation of Radar Reflectivity Data in Convective Precipitation Forecasting

Scale Compatibility Analysis in Geographic Process Research: A Case Study of a Meteorological Simulation in Hong Kong

Sampling and misspecification errors in the estimation of observation‐error covariance matrices using observation‐minus‐background and observation‐minus‐analysis statistics

Impact of Multivariate Background Error Covariance on the WRF-3DVAR Assimilation for the Yellow Sea Fog Modeling

Multivariate bias correction of regional climate model boundary conditions

The Importance of Data Assimilation Components for Initial Conditions and Subsequent Error Growth

Improving the Gaussianity of Radar Reflectivity Departures between Observations and Simulations by Using the Symmetric Rain Rate

Assimilation of the deep learning-corrected global forecast system fields into the regional model for improving medium-range persistent precipitation forecasts

Modifying Covariance Localization to Mitigate Sampling Errors from the Ensemble Data Assimilation

How much improvement can precipitation data fusion achieve with a Multiscale Kalman Smoother-based framework?

Construction and Characteristic Analysis of Background Error Covariance Coupled with Land Surface Temperature

Scale Matching of Multiscale Digital Elevation Model (DEM) Data and the Weather Research and Forecasting (WRF) Model: a Case Study of Meteorological Simulation in Hong Kong