SGER: Performance-based Probabilistic Multi-Model Climate Change Scenarios

这项研究正在解决以下问题：人为强迫可能如何影响区域气候？有什么把握？到什么程度？本工作主要包括两个部分：(1)利用用于21世纪气候变化预测的大气-海洋全球环流模式（AOGCM）验证20世纪时间特征的表现，如多年代际平均值的趋势和这些平均值的年际变率；(2)构建21世纪气候及其变率的概率多模式情景，将(1)的结果作为模式预测权重分配的客观依据。对于验证分析，将对AOGCM集合应用概率技能度量，相对于合成观测集合，每个集合都具有相同的20世纪观测气候的时间特征。类似地，观察结果的子集将用于构建期望技能分数的概率分布。对于aogcm的蒙特卡洛技能分数的概率分布与观测值显著相似的区域，将认为20世纪的模型性能是可信的。对于概率型多模式变化情景，将采用贝叶斯方法，使用先前的假设，即21世纪的变率可能由观测到的20世纪变率表示。对季节性（即3个月平均值）近地表气温和降水进行从网格尺度到区域尺度的空间分析。本研究结果拟纳入政府间气候变化专门委员会（IPCC）第四次评估报告。本研究采用的方法是基于模式性能的多模式集成技术，这些方法尚未在气候变化预测中得到应用，可以为未来的IPCC报告进行进一步研究。这项研究的结果还可用于为季节至年际气候变率和预测设定更长期的背景。这里的建议是，适当地综合季节性预测和长期评估，考虑到每一个的不确定性，提供了最大限度地减少损失的最佳机会，利用机会，并朝着可持续的做法努力。更广泛的影响包括培养博士后，以及对气候预测模型性能的可靠估计所带来的社会效益。

This research is addressing the questions: How is anthropogenic forcing likely to affect regional climate? With what certainty? To what degree? The work consists of two main parts: (1) Verification of the performance of 20th Century temporal characteristics, such as the trends in multi-decadal means and the interannual variability about those means, from the Atmosphere-Ocean Global Circulation Models (AOGCM)s used for 21st Century climate change predictions; and, (2) Construction of probabilistic multi-model scenarios for 21st Century climate and its variability, using the results from (1) as an objective basis for assigning weights to the model predictions. For the verification analysis, a probabilistic skill metric will be applied to the AOGCM ensembles relative to an ensemble of synthetic observations, each of which possesses the same temporal characteristics of the observed climate over the 20th Century. Similarly, sub-sets of the observations will be used to construct the probability distributions of the expected skill score. For regions where the probability distributions of the Monte Carlo skill scores for the AOGCMs and observations are significantly similar, the model performance for the 20th Century will be deemed credible. For the probabilistic multimodel change scenarios, a Bayesian approach will be applied using the prior assumption that 21st Century variability may be represented by the observed 20th Century variability. The analyses will be performed spatially from the grid scale to the regional scale for seasonal (i.e. 3-month mean) near-surface air temperature and precipitation. The results from this research are intended for inclusion in the Fourth Assessment Report of the Intergovernmental Panel for Climate Change (IPCC). The methods employed in this research, applying techniques of multi-model ensembling that are based on model performance, have not yet been applied in the context of climate change predictions and can pilot further research for future IPCC reports. The findings of this research also can be used to set the longer-range context for seasonal-to-interannual climate variability and predictions. The suggestion here is that proper synthesis of seasonal forecasts, and longer-term assessments, taking into account the uncertainties of each, provides the best opportunity to minimize losses, take advantage of opportunity, and work toward sustainable practices. Broader impacts include the training of a post-doc and the societal benefits to be realized from credible estimates of the performance of climate prediction models.