CMIP5气候模式下淡水通量变化

基于全球降水气候态计划(GPCP)的降水资料和美国伍兹霍尔海洋研究所(WHOI)的客观分析海气通量(OAFlux)的蒸发数据,对CMIP5的13个耦合模式的淡水通量历史模拟结果进行评估。结果表明:模式能够模拟出淡水通量的气候态空间分布,但普遍存在双热带辐合带(ITCZ)现象,热带海域是模式模拟不确定性最大的区域。模式能较好模拟出纬向平均的淡水通量的分布特征,但量值较实测偏小,且由于模式对1月10°S附近淡水通量的模拟过低,导致年平均的赤道和10°S之间的淡水通量模拟存在明显的偏差。季节尺度上,模式对北半球淡水通量的变化特征有很好的模拟能力,但对南半球的模拟能力不足。年际尺度上,模式普遍能够刻画ENSO引起的淡水通量在太平洋中部同西太平洋以及印尼贯通流反相变化的空间分布特征,但是时间特征模拟很差。从各个方面评估模式的历史模拟结果,多模式集合的结果都要优于单个模式的结果。全球变暖背景下,未来淡水通量变化最显著的区域位于热带和亚热带区域。原本蒸发(降水)占主导的海域,蒸发(降水)更强。不同气候情景下,淡水通量变化的空间形态没有显著变化,但RCP8.5气候情景下模拟的淡水通量变化幅度及模式间变化的一致性均强于RCP4.5的结果。

Freshwater flux variations based on CMIP5 climate models

Reanalysis precipitation datasets from Global Precipitation Climatology Project (GPCP) and evaporation datasets from WHOI OAflux project are used to evaluate the historical freshwater flux from 13 CMIP5 climate models. The results show that all models could capture the climatological space distribution while the double ITCZs are widespread and the tropical ocean is the region with most significant uncertainties among the models. Latitudinal distributions of zonal-mean freshwater and their seasonal variations are broadly captured by most of the models. But the models' results are smaller than the reanalysis data,freshwater simulation between 10°S and the equator shows significant bias because of the overestimate of freshwater in the January. On the seasonal timescale,the models show good result in the North Hemisphere,but have some deficiencies in the South Hemisphere. On the interannual timescale,the models could provide a suitable simulation of freshwater space distribution induced by ENSO,but they are deficient in simulating the temporal characteristics. Results of multi-model ensemble are superior to single model in all respects of evaluation.Tropical and sub-tropical ocean will change significant under the circumstance of global warming,where used to be dry will be drier and wet will be wetter. Spatial distribution pattern will not change markedly under different scenarios,but the amplitudes of freshwater flux and the consistency among the models will strengthen under RCP8.5 compared with the RCP4.5.