【作者】 朱献；

【导师】 董文杰；

【作者基本信息】 兰州大学 ， 气象学， 2013， 硕士

【摘要】 全球变暖已经是一个不争的事实,并且全球变暖的归因问题是当今全球变化和地球系统科学无法回避的基本科学问题。当前国际关于气温、降水等要素的归因研究已经开展很多,然而关于北半球积雪的归因研究开展很少。本文分析了北半球积雪面积的时空变化特征,评估了参加CMIP5的耦合气候模式对北半球春季积雪面积及北极涛动(AO)的模拟能力并与CMIP3模式结果进行了比较。应用多模式集合对未来不同排放情景下北半球春季积雪面积进行了预估。分析了海平面气压、北极涛动与北半球春季积雪面积的关系,不同强迫对北半球春季积雪面积的影响。对比了所有强迫、自然强迫、温室气体强迫三组历史模拟试验下CMIP5耦合模式模拟的北半球春季积雪面积及海平面气压、北极涛动的变化特征。基于三组历史试验结果,应用统计分析方法对北半球春季积雪面积和北极涛动进行了归因分析。主要得到以下结论：(1)北半球积雪面积的显著减少主要发生在春季和夏季,而在增温最显著的冬季和秋季减少不明显,所以温度并不是制约积雪存在的唯一主要因素。北半球积雪面积变化的关键区是以青藏高原为代表的中纬度地区,过去几十年中纬度地区的积雪减少最明显。不同季节北半球积雪变化的周期特征不同。冬季变化周期表现为7.8年以及3-4年,秋季表现出4-5年的变化周期；春季和夏季北半球积雪以周期为8年左右的变化为主。(2) CMIP5耦合模式对北半球积雪面积具有一定的模拟能力,整体上能模拟出积雪的分布特征,模式模拟的积雪面积与观测值的空间相关系数均达到了0.9。多数模式低估了北半球积雪面积的减少趋势,多模式集合对模拟效果有一定的改进。对比CMIP3模式,CMIP5模式对北半球积雪的模拟有一定的改进。未来不同温室气体排放情景下,应用23个CMIP5模式集合平均对未来北半球春季积雪面积预估发现,未来不同温室气体排放情景下北半球积雪面积变化显著不同。在低排放情景下,积雪面积减少不明显；而在高排放情景下,积雪面积将显著减少。整体上,温室气体排放越多,未来北半球积雪面积减少越明显,控制温室气体排放对于未来北半球积雪的生存至关重要。(3) CMIP5模式能够较好地模拟出冬、春季AO模态的空间分布特征,模拟的AO模态与观测的空间相关系数均超过了0.6,并且标准差也接近于观测。尽管CMIP5模式没能模拟出冬、春季AO指数前30年处于显著负位相后20年处于显著的正位相的显著特征,但是基本能够模拟出冬、春季AO指数在50年间表现出的显著的增强趋势以及周期特征,多模式集合改进了模拟效果。尽管CMIP5模式对冬、春季AO指数的模拟能力还不够理想,没有完全模拟出AO指数的变化特征,但对比CMIP3模式结果,无论是对AO模态还是对AO指数的模拟都有一定改进。(4)北半球春季积雪面积与冬、春季北极涛动呈现出负相关关系,并且积雪面积与冬季北极涛动的负相关要更显著。前冬海平面气压变化造成北极涛动的增强从而加快北半球春季积雪面积的减少。前冬北极涛动的显著增强可能是造成北半球春季积雪面积减少的一个重要因素。(5)不同强迫与北半球春季积雪面积的相关性差异显著,温室气体强迫与北半球春季积雪面积有很强的正相关,自然强迫与北半球春季积雪面积的相关性不显著。并且不同强迫对北半球积雪面积的影响显著不同。温室气体强迫对北半球积雪面积的减少起到显著的促进作用,而其它强迫则显著的抑制了北半球积雪面积的减少。不同外强迫下CMIP5模式的历史模拟试验表明,北半球春季积雪的减少主要发生在1950年后,在温室气体强迫试验中北半球春季积雪减少最为显著,在自然强迫试验中没有表现出减少趋势。由所有强迫和温室气体强迫试验得到的海平面气压所表现出来的趋势分布特征与观测的一致,并且强度上温室气体强迫试验更为接近于观测,自然强迫试验得到的海平面气压的趋势特征则相反。同样温室气体强迫试验得到的北极涛动最为接近于观测。基于三组历史试验结果,应用统计方法分析发现北半球春季积雪面积的减少以及北极涛动的增强主要是温室气体强迫造成的。更多还原

【Abstract】 Global warming is an indisputable fact. The attribution of global warming is a basic question of Global Change and Earth System Science. In the current, there are lots of attribution studies for temperature, precipitation and so on, but the research which is for the attribution of Northern Hemisphere snow cover is less. In this study, the temporal and spatial distributions of Northern Hemisphere snow cover were analyzed, and the simulating capability of snow cover and Arctic Oscillation (AO) of global climate models which joined the CMIP5were tested, and the results were compared with CMIP3models. Then we used the method of model ensemble average to predict the variation of Northern Hemisphere snow cover in future in RCP2.6, RCP4.5, RCP6.0and RCP8.5. The connection which existing between snow cover and sea level pressure, AO in Northern Hemisphere and the influence which radiative forcings produced on snow cover were analysised. The changes of snow cover, sea level pressure and AO which simulated by coupled models were contrasted in Historical, HistoricalNat and HistoricalGHG scenarios. Based on the results of three historical experiments, attribution of Northern Hemisphere spring snow cover and AO are analyzed. The main contents and conclusions are as the follow:(1)The significant decrease of Northern Hemisphere snow cover occurs mainly in the spring and summer, not in winter and autumn, although global warming is the most significant in winter. So, the temperature is not the only major factor that restricted the snow cover. The key area where the reduction of snow cover is the most obvious over the past few decadal is Qinghai-Tibet Plateau in the mid-latitude region. Cycle characteristics of snow cover are different in four seasons. There are mainly high frequency changes of snow cover in winter and autumn with the cycles of7-8years,3-4years in winter and4-5years in autumn, and the8years cycles of high-frequency change is represented in spring and summer.(2)The CMIP5models simulating capability of Northern Hemisphere snow cover is well. Overall the distribution characteristics of snow cover can be presented in the simulation by CMIP5models with the spatial correlations which existing between simulations and observation are more then0.9. But the simulation of snow cover in complex terrain area for example Qinghai-Tibet Plateau is poor and the significant trend towards a reduced spring snow cover extent over the1979-2005is underestimated. The multi-model ensemble has improved the effect of the simulations by CMIP5models. Through comparison, the simulation of snow cover by CMIP5models is better than CMIP3models. The projection by multi-model ensemble show that the changes of snow cover under different greenhouse gas emission scenarios. The northern Hemisphere snow cover will reduce most significantly and continue until the end of the century under RCP8.5scenario; Under RCP2.6, the snow cover will reduce in the first half of this century and remain stable after2040. Therefore, the more greenhouse gas emissions the more obviously Northern Hemisphere snow cover will reduce. It is crucial to control the discharge of GHG emissions for mitigating the disappearance of snow cover over Northern Hemisphere.(3)The basic characteristics of winter and spring AO modal can be reproduced by CMIP5models. A strong positive spatial correlation exits between the AO modal simulated by models and the observed with the correlation coefficient reaching0.6and the standard deviation of the AO modal simulated is close to the observed. The significant increasing trend and oscillation cycle of winter and spring AO index can be reproduced by models although the salient features that winter and spring AO is significant negative phase in the first30years and positive phase in the last20years aren’t shown up in the simulation. The multi-model ensemble has improved the effect of the simulations by CMIP5models. Although the simulations aren’t good enough to catch all the significant features of winter and spring AO index by CMIP5models, but relative to CMIP3models, there are significant improvement not only for the reducing trend but also for the change cycle in the simulations by CMIP5models.(4)The negative relationship exists between Northern Hemisphere spring snow cover and winter or spring AO with the relationship that existing between snow cover and winter AO is stronger. The enhancement of winter AO caused by the change of preceding winter SLP can produce significant impact on the Northern Hemisphere snow cover, the enhancement of the previous winter AO will accelerate the reduction of the Northern Hemisphere spring snow cover. The enhancement of winter AO is one of the important factors that have cause the reduction of Northern Hemisphere spring snow cover.(5)The correlations between Northern Hemisphere spring snow cover and the different radiative forcings are significantly different. A strong negative correlation exists between SCA and GHGs forcing. The correlation between SCA and natural forcing or that between SCA and precipitation is not obvious. There are different roles that different forcings have played in snow cover. The GHGs forcing played significant promoting effect in the reduction of snow cover, while the other forcing played a significant inhibitory effect; however, the effect natural forcing played is not significant. The historical simulation experiments with different forcings in CMIP5show that, the decrease of snow cover mainly occurred after1950, and in the HistoricalGHG experiment the decrease trend is the most obvious, there isn’t decrease trend in the HistoricalNat experiment. In the Historical and HistoricalGHG experiments, the distribution characteristics of the trend in sea level pressure is consistent with observation, and that in HistoricalGHG is more close to observation. There is opposite characteristics presented in the sea level pressure which simulated in HistoricalNat experiment. The same as sea level pressure, AO simulated in the HistoricalGHG experiment is the most close to observation. Based on the results of three historical experiments, statistical analysis showed that the decrease of Northern Hemisphere spring snow cover and the enhancement of winter AO were due to the Greenhouse gases.更多还原