Thepaperinvestigatestheabilitytoretrievethetruesoilmoistureprofilebyassimilatingnear-surfacesoilmoistureintoasoilmoisturemodelwithanensembleKalmanfilter(EnKF)assimilationscheme,includingtheeffectofensemblesize,updateintervalandnonlinearitiesintheprofileretrieval,therequiredtimeforfullretrievalofthesoilmoistureprofiles,andthepossibleinfluenceofthedepthofthesoilmoistureobservation.Thesequestionsareaddressedbyadesktopstudyusingsyntheticdata.The'true'soilmoistureprofilesaregeneratedfromthesoilmoisturemodelundertheboundaryconditionof0.5cmd-1evaporation.Totesttheassimilationschemes,themodelisinitializedwithapoorinitialguessofthesoilmoistureprofile,anddifferentensemblesizesaretestedshowingthatanensembleof40membersisenoughtorepresentthecovarianceofthemodelforecasts.Alsocomparedaretheresultswiththosefromthedirectinsertionassimilationscheme,showingthattheEnKFissuperiortothedirectinsertionassimilationscheme,forhourlyobservations,withretrievalofthesoilmoistureprofilebeingachievedin16hascomparedto12daysormore.Fordailyobservations,thetruesoilmoistureprofileisachievedinabout15dayswiththeEnKF,butitisimpossibletoapproximatethetruemoisturewithin18daysbyusingdirectinsertion.ItisalsofoundthatobservationdepthdoesnothaveasignificanteffectonprofileretrievaltimefortheEnKF.Thenonlinearitieshavesomenegativeinfluenceontheoptimalestimatesofsoilmoistureprofilebutnotveryseriously.
