基于标准化时间序列加权法的IPCC-AR4多模式集成

采用对标准化时间序列作加权的方法做多模式集成,并以相关系数与均方根误差作为衡量模式模拟预估能力的标准,将IPCC-AR4的 10个气候模式模拟1961-1999年(39a)中国降水的标准化加权集成结果与气候模式修正、Taylor函数加权集成结果进行比较,结果显示前者具有明显优势。为进一步验证这种优化的性能是否具备稳定性,将由控制阶段(1961-1999年)模式与实测数据计算得到的权重系数应用于A1B排放情景下的模式降水数据,预估研究区域2000-2012年(13a)的降水,并与同时间段的模式修正、Taylor函数加权集成进行对比。结果表明,该集成结果在预估阶段(2000-2012年)仍优于模式修正、Taylor函数加权集成结果。

Multi-model integration of IPCC-AR4 based on normalized time sequence weighting method

Through adopting the standardized time sequence weightening method to carry out multi-model integration, and taking the relative coeffients and root mean square error as the measuring standard for the evaluation ability of models, the simulated precipitation by normalized ensemble of 10 models in the IPCC-AR4 during 1961-1999 over China are compared with the results obtained by fit revising method and Taylor function. Results show that the former is much better than the latter two methods. To verify whether the capability of the optimization obtained from the simulations can be maintained in the projections, the weights fitted from the simulations are used to ensemble the projected precipitations during 2000-2012 under the A1B emission scenario, which demonstrates that the projection skill is improved more by the ensemble method than by the other two methods.