利用Downscaled CMIP3 and CMIP5 Climate and Hydrology Projections(DCHP)提供的31个CMIP5降尺度数据和CRU逐月气温、降水格点数据集,通过评估PLS(偏最小二乘回归)、RR(岭回归)和EE(等权平均) 3种多模式集合平均预估模型对历史气候变化的模拟能力,确定最优集合方法,进而预估开都-孔雀河流域21世纪气候变化情景。结果表明: ①所建立的PLS模型对流域的气温和降水具有较好的模拟能力,尤其对气温的模拟,r值均达到了0.64以上,明显优于降水(0.19~ 0.36),但存在空间异质性; ② 21世纪开都-孔雀河流域各子区气温呈显著增加趋势,且RCP8.5情景下的增温速率〔0.58~ 0.67 ℃·(10a)~(-1)〕是RCP4.5情景下〔0.25~ 0.31 ℃·(10a)~(-1)〕的2倍以上, 21世纪中叶是2种情景产生明显差异的开始。整个流域增温速率由西北山区向东南荒漠区逐渐增大; ③未来降水在不同排放情景下变化速率的分布状况略有不同,但均呈显著增加趋势,且RCP8.5情景下的增加速率〔1.22%~ 1.54%·(10a)~(-1)〕总体上高于RCP4.5〔0.80%~ 1.32%·(10a)~(-1)〕。
英文摘要:
Climate change assessments on both global and regional scales rely strongly on the global climate models (GCMs) which are dominantly provided by the Coupled Model Inter Comparison Project Phase 5 (CMIP5). Based on the grid datasets of monthly air temperature and precipitation from CRU (Climate Research Unit) and 31 CMIP5 GCMS data from the Downscaled CMIP3 and CMIP5 Climate and Hydrology Projections (DCHP),in this paper the performance of three Multi-Model Ensemble Mean methods (PLS,RR and EE) in simulating the historical climate change processes was evaluated,and the optimal ensemble method was determined and estimated for predicting the future climate change in the Kaidu-Konqi River Basin in the 21st century. The results indicated that: ① The performance of the established Partial Least Squares (PLS) model was the best in simulating air temperature and precipitation in the study area. The r values of simulated temperature were all higher than 0.64, they were obviously better than those of simulated precipitation (0.19 - 0.36). However, there was a spatial heterogeneity in both temperature and precipitation simulations; ② In the 21st century, the air temperature in the 4 sub-basins of Kaidu- Kongqi River Basin would be in a significant increase trend. The increase rates of air temperature 〔0.58 - 0.67 ℃·(10a)~(-1)〕under the RCP8.5 scenario would be doubled compared with those under RCP4.5 scenario 〔0.25-0.31 ℃·(10a)~(-1)〕. The significant difference between the two scenarios would begin from the mid-21st century. From the perspective of entire watershed, the warming rate increased gradually from the mountainous area in the northwest to the desert in the southeast; ③ The distribution of change rates of precipitation was slightly different under different discharge scenarios,but both of them would be in a significant increase trend. The increase rate under RCP8.5 scenario〔1.22%-1.54%·(10a)~(-1)〕would be holistically higher than that under RCP4.5 scenario 〔0.80%-1.32%·(10a)~(-1)〕.