Assessment of Unforced Variability as a Source of Uncertainty in Regional Hydroclimate Simulations and Projections

嵌套区域气候模式（RCMs）是全球范围内用于产生气候变化预估和适合水文研究尺度的长期模拟的主要工具。需要资金来进行我的创新的非定向研究，以调查对估计区域气候预测和水管理模拟的不确定性至关重要的问题。一般的看法是，与与排放情景或GCM缺陷相关的不确定性相比，GCM或RCM内部变率（IV）对总体不确定性的影响较小。然而，De Wit等人（2007）指出，受同一GCM强迫的不同rcm会产生不同的平均季节性降水和蒸发模式，Horton等人（2006）发现，给定排放情景下的模拟排放显示出由驱动GCM和驱动rcm两种因素引起的较大变异性。这些研究表明，由相同GCM驱动的RCM集合可以产生与不同GCM驱动的相同RCM所产生的变率相同量级的水相关变量的变率。最近，Musíc和Caya（正在准备中）表明，在流域尺度上，GCM IV对径流不确定性的贡献可以与使用不同的GCM所带来的贡献一样大。Braun et al.（2011）表明，在流域尺度的径流分析中，使用大域的RCM中的IV可以达到接近GCM IV的值。因此，在每一个使用rcm的水文研究中，都必须考虑gcm和rcm的非强迫变率。以上研究均忽略了非强迫变率的贡献。本文旨在估计非强迫变率的空间分布及其对总体不确定性的相对贡献。由Ouranos（在我的领导下）生成的用于解决这种非强制变异性分析的集成现在已经完成并且可用。将利用最近获得的区域模拟集合对非强迫变率进行全面分析。

Nested Regional Climate Models (RCMs) are the main tool used worldwide to generate climate change projections and long simulations at scales suitable for hydrological studies. Funds to pursue my innovative non-directed research are required to investigate issues that are vital for the estimation of uncertainty in regional climate projections and simulations for water management. The general perception is that GCM or RCM internal variabilities (IV) are minor contributors to the overall uncertainty in comparison to uncertainty associated to emissions scenarios or to deficiencies in GCMs. However, De Wit et al. (2007) stated that different RCMs forced by the same GCM generate different mean seasonal precipitation and evaporation patterns and Horton et al. (2006) found that the simulated discharges for a given emissions scenario show large variability induced by both, the driving GCMs and the driven RCMs. These studies demonstrated that a RCM ensemble forced by the same GCM could generate variability in water related variables of the same magnitude as the variability produced by using the same RCM driven by different GCMs. Recently, Musíc and Caya (in preparation) showed that the GCM IV contribution to uncertainty in runoff at the watershed scale can be as large as the one introduced by using a different GCM. Braun et al. (2011) showed that the IV in RCM using large domains could reach values close the GCM IV in analysis of runoff made at the watershed scale. Therefore, the unforced variability in both GCMs and RCMs has to be taken into account in every hydrology studies using RCMs. All of the above studies neglected the contribution from the unforced variability. The present proposal wants to estimate the spatial distribution of the unforced variability and its relative contribution to the overall uncertainty. The ensemble generated by Ouranos (under my leadership) that was designed to address this unforced variability analysis is now complete and available. A complete analysis of the unforced variability will be conducted using this recently available ensemble of regional simulations.