Bystatisticanddynamicanalyses,wehavecometothefollowingconclusions:(1)TheECMWFmedium-termnumericalforecastcanforecastmedium-termactivityofsubtropicalhigh,andtheaccuracyrateofforecastcannothavelargeimprovementbytranslationalcorrections.(2)TheimportantcausefortheECMWFmedium-termnumericalforecasttohaveerrorsin1998isthattheastronomicaltideisnotincludedinthemodel.(3)TwoindexesarefoundfromwhichitcanbejudgedthatECMWFmedium-termnumericalforecastwillhaveerrorsiftheastronomicaltideisignoredinthemodel:①Whenthe54.7°lineunderthemoonofthenodicalmonthastronomicalsingularitiescoincideswiththetrough-lineofthesubtropicaljetflowfrom50°Eto150°Eonthe500hPalevelat2000L.T.ofthesameday,andisapproximatelyvertical(α>60°)withtheisotherm,thentheday0-2daysaftertheappearanceofthenodicalmonthastronomicalsingularitiesisdefinedastheinitialday.Theninthreesuccessivedaysaftertheinitialday,ECMWFmedium-termnumericalforecastofthenorthernlatitudeofthe588lineat120°Ewillhavecontinuouserrorsaslargeastwolatitudes(7/9).Otherwise,itwon'thavecontinuouserrors(13/13).②Otherwise,ifthe54.7°lineisintherangeofalowpressurebetweentwohighpressures,thenthereisadispersiveerroronthedayofthenodicalmonthastronomicalsingularities(5/7).Thereisnotanyerror(6/6)otherwise.
