Inthispaper,anewbiasestimationmethodisproposedandappliedinaregionalensembleKalmanfilter(EnKF)basedontheWeatherResearchandForecasting(WRF)Model.Themethodisbasedonahomogeneouslinearbiasmodel,andthemodelbiasisestimatedusingstatisticsateachassimilationcycle,whichisdifferentfromthestateaugmentationmethodsproposedinpreviousliteratures.Thenewmethodprovidesagoodestimationforthemodelbiasofsomespecificvariables,suchassealevelpressure(SLP).AseriesofnumericalexperimentswithEnKFareperformedtoexaminethenewmethodunderasevereweathercondition.Resultsshowthepositiveeffectofthemethodontheforecastingofcirculationpatternandmeso-scalesystems,andthereductionofanalysiserrors.ThebackgrounderrorcovariancestructuresofsurfacevariablesandtheeffectsofmodelsystembiasonEnKFarealsostudiedundertheerrorcovariancestructuresandanewconcept‘correlationscale’isintroduced.However,thenewmethodneedsfurtherevaluationwithmorecasesofassimilation.
