This paperinvestigatesamodifiedgreymodelforforecastingtheinflowofareservoir.Theintegral form ofthebackgroundvalueisemployedfortheoriginalgreymodel,GM(1,1),toimproveaccuracy and applicability.Thereafter,theFourierseriesisalteredtohandleextremevalueswithregardto prediction;exponentialsmoothingisusedtoimprovethedrawbacksofthepredictiondelay phenomenon.Finally,wearehybridisedastheultimategreymodelwithoutstandingprediction accuracy,namelyEFGM(1,1).Asatyphooncausessignificantchangesintheinflowofareservoir,this paper appliesthefuzzymembershipfunctionfordealingwithitduringthefloodseasontoconstruct the fuzzygreymodificationmodel,FEFGM(1,1).Resultsofgreymodelsarecomparedwiththoseofthe Autoregressive Integrated Moving Average(ARIMA). By evaluatin g different indices, thee r rors of the predictedextremevalueofEFGM(1,1)performbetterthanthoseofGM(1,1)andARIMA,however worse thanthatofFEFGM(1,1).ThefinalFEFGM(1,1)showshighprecisionwithregardtoreservoir inflow predictionduringtyphoonswithcombinedeffectsoffuzzy,exponentialsmoothing,Fourier series