Intercomparison And Evaluation Between Grid Precipitation Data And Observations In Dongting Lake Basin

Accurate precipitation data is important in many applications,such as agricultural production,water resource planning and management,flood and drought monitoring,geological disaster warning,etc.The traditional way to obtain precipitation data is from ground station observation.With the development of technology,more and more precipitation data were produced based on numerical model,interpolation of gauge observation,satellite retrieval,data combining.The validation and evaluation of grid precipitation data has been an intense topic of research to meteorological researchers.The dissertation was performed to validate the grid precipitation data in Dongting lake basin(107°13'~114°18' E?24°35?~30°27? N).The grid data were from CFSR(NCEP Climate Forecast System Reanalysis),ERA-Interim(European Centre for Medium-Range Weather Forecasts),CMAP(the Climate Prediction Center Merged Analysis of Precipitation),GPCP(Global Precipitation Climatology Project),CRU(Climatic Research Unit,University of East Anglia)and PREC/L(NOAA's Precipitation Reconstruction over Land).Based on the observations of27 weather stations from CMD(China Meteorological Data),the difference in spatial and temporal distribution,different relationship between precipitation and the monsoon,Western Pacific Subtropical High,ENSO were compared.The variation characteristics of flood and drought were studied too.The objective was to find out the most suitable data for Dongting lake basin by assessing the performance of different precipitation data.The main results were as follows.(1)The spatial distribution characteristics of annual precipitation in Dongting Lake basin appeared on a diminishing scale from east to west and from south to north.The spatial distribution of precipitation difference by warm and cold season was similarly to that of annual precipitation,and so did the precipitation in June.CFSR and ERA-Interim data overestimated the precipitation in most areas and could partly reproduce the spatial distribution of CMD annual precipitation.The coexistence of overestimation and underestimation were found in CMAP,GPCP,CRU and PREC/L,which spatial distributions of seasonal difference and monthly precipitation in some regions were consistent with those of CMD.PREC/L precipitation showed higher consistency with CMD in space.The spatial distributions of EOF1 in all grid data and EOF2 in CMAP and GPCP were similar to that of CMD,while the spatial distribution of EOF3 in most grid data was quite different.(2)The annual precipitation in Dongting lake basin was characterized by the change of pluvial-dry-pluvial,and no significant changes have occurred.The change trend in grid precipitation data was consistent with that of CMD.There was more rainfall in spring,summer in Dongting Lake basin,but less in autumn and winter.The spatial distribution of monthly precipitation in CMD could be well represented by all grid data.CMAP and ERA-Interim overestimated most monthly precipitation,especially in summer months;CMAP and CRU underestimated most monthly precipitation;GPCP and PREC/L overestimated precipitation in warm season but underestimated in cold season.GPCP data agreed well with CMD in monthly scale.The first two change periods on annual precipitation in grid data were similar to those in CMD,but the latter two periods differed greatly.(3)There were significantly negative correlation between CMD precipitation in July and South Asian summer monsoon index II(SASMI-II),which was represented by CFSR?CMAP?GPCP?CRU and PREC/L.The negative correlation between CMD precipitation in August and Asian summer monsoon index(ASMI)was also significant,so did to CFSR?CMAP?GPCP and CRU.The spatial distribution of correlation coefficient was similar in grid data and CMD.The number of station with same period of IMF3 components in CMD July precipitation as June SASMI-II corresponding IMF were the most,which the ratio is about 50%,and so did in CMD August precipitation and ASMI.From the space distribution,the IMF3 period in June could be represented such as Dongting lake Plain and central-southern part of the basin in most grid data.In August,it was well done in the middle reach of the Xiang River,edge of the mountain regions in the southeast of the basin.(4)All the grid data could reflect the influence caused by ridge line in July,northern boundary in June and July,west-extending ridge point position from 6~8on CMD preciptation data.To the grid data,the fitting ability of five indices of WPSH in partial area were close to that of CMD in June and July.In August,the fitting ability had a good consistance with that of CMD in the whole basin for ERA-Interim.(5)There were significantly positive correlation between IMF4 and trend of annual precipitation in CMD and mean annual SSTA of Ni?o3.4,which could be reflected by grid data except for CFSR.In CMD,the significant correlation coefficient between IMF4 and mean annual SSTA extended most area of Donting lake basin and it was well presented in northwestern and southern by CMAP?PREC/L,in northwestern by GPCP.The postive relation between monthly SSTAand precipitation lagging 6 months was the largest during warm episode.The correlation coefficient in CMD and grid data decreased from northwest to Southeast in space.The postive relation between monthly SSTA and precipitation in the same time was the largest during clod episode.For most grid data,the higher correlation coefficient in southern area was consistance with CMD.The area of significant correlation coefficient in grid data was larger than CMD.(6)Among the drought and flood grades determined by the Z index,the years of heavy flood?flood?heavy drought?drought in GPCP agreed best with those in CMD.The strongest reproduction ablity to spring flood and drought were from CMAP and PREC/L,summer flood and drought from ERA-Interim and GPCP.The worst were from CFSR and CRU respectively.The yearly drought frequency in western and eastern,spring drought frequency in northwestern and southern from most grid data,the summer drought frequency in southwestern and northwestern from all grid data were similar to CMD.The annual flood frequency in western and southwestern,the spring flood frequency in southwestern and northwestern,the summer flood frequency in northwestern and southern could be represented in some grid data.There were more stations in GPCP which had the same annual drought and flood frequency as in CMD,so did summer drought and flood frequency in PREC/L,spring drought frequency in CRU,spring flood frequency in ERA-Interim.