Research On The Error And Predictability Of Climate Model

Given observation SST and climate SST(average among hindcasting period), T63L16 GCM model is respectively used to perform numerical experiment of ensemble hindcasting for twenty summer,from 1982 to 2001, then systematical test results of globe and East Asian scale. Basing on those,error evolvement mechanism, atmosphere predictability,and initial atmosphere anomalies impacts on seasonal prediction and other aspect are elementary discussed,getting some interesting results. Primary result are sum up below:(1) This model can preferably simulates summer average circulation of globe scale, well and truly predicts activity center on high, midst and low troposphere of atmosphere circulation, but among different weather climate systems, their location and intensity have some error. For temperature and precipitation, this model has some simulation ability, but results great away from practicality observation data.(2) The precipitation of East area from this model is distinctness excessive,and there exists one nonexistent precipitation center on east Tibet plateau. The error of convective-precipitation makes up to ninety percent of total precipitation error , but large-scale precipitation only very small percents.And results from cloud cover, surface soil water, surface sensible heat flux and surface latent heat flux indicate that their have some effects on local precipitation simulation.(3) Systematic error and random error have the same order of magnitude, but random error is slight more than systematic error. From the error equation, the factors dominate evolvement of systematic error and random error are convergenc-divergence of flux term, mutual conversion term, barotropic generation term and source-sink term. But their effect are different from different simulation phase.The mechanism of evolvement of systematic error and random error aren't sameness, the most discrepancy is that mix-generation term accelerates systematic error development the whole simulation period, but holds back for random error. There exists conversion between systematic error and random error, and among conversion term, latitude wind systematic error is dominating.(4) Research of the predictability of sea mean pressure, 500hPa geopotential height, precipitation, 850hPa temperature and surface soil indicate predictability of sea mean pressure fastens on tropic, low predictability is extratropical area. Predictability of 500hPa geopotential height locates tropic, exhibiting belt form. The range of predictability of precipitation is narrow. Predictability of temperature is on tropic Pacific ocean area, and some area of north American has some predictability. Predictability of soil is on mostly land of east Asian and north American.Taking the 500hPa geopotential height for example,analysising the predictability from different climate backgrounds indicate, predictability in ENSO year is 0.2 more than other backgrounds on tropic .Predictability on extratropic of the northern hemisphere is smaller than others. Predictability from ENSO year has the same shape from not ENSO but greatly different from not ENSO year.(5) Initial atmosphere anomaly has greatly effect on summer seasonal predict and this effect varies among different variables. To sea mean pressure, greatly influence are Atlantic and south Pacific ocean.Impact on geopotential height most covers north Pacific ocean and Atlantic, but area is wider than sea mean pressure, and its effect on the top of troposphere are greater than of midst and low. Impact on wind is smaller than the former two variables. Effect of initial atmosphere anomaly on east Asian is morer than globe average. Effect on geopotential height is even evident in the year climate varied greatly. Result from typical years shows that model can successfully simulate interannual change of wind direction on tropic troposphere and stratosphere , proving that the effect of Initial atmosphere anomaly is even distinctness in typical year, and interannual change of summer latituide wind on tropic troposphere and stratosphere incl...