Abstract Climate change impact assessment on meteorological variables involves large uncertainties as a result of incomplete knowledge on the future greenhouse gas concentrations and climate model physics, next to the inherent internal variability of the climate system. Given that the alteration in greenhouse gas concentrations is the driver for the change, one expects the impacts to be highly dependent on the considered greenhouse gas scenario (GHS). In this study, we denote this behavior as GHS sensitivity. Due to the climate model related uncertainties, this sensitivity is, at local scale, not always that strong as expected. This paper aims to study the GHS sensitivity and its contributing role to climate scenarios for a case study in Belgium. An ensemble of 160 CMIP5 climate model runs is considered and climate change signals are studied for precipitation accumulation, daily precipitation intensities and wet day frequencies. This was done for the different seasons of the year and the scenario periods 2011–2040, 2031–2060, 2051–2081 and 2071–2100. By means of variance decomposition, the total variance in the climate change signals was separated in the contribution of the differences in GHSs and the other model-related uncertainty sources. These contributions were found dependent on the variable and season. Following the time of emergence concept, the GHS uncertainty contribution is found dependent on the time horizon and increases over time. For the most distinct time horizon (2071–2100), the climate model uncertainty accounts for the largest uncertainty contribution. The GHS differences explain up to 18% of the total variance in the climate change signals. The results point further at the importance of the climate model ensemble design, specifically the ensemble size and the combination of climate models, whereupon climate scenarios are based. The numerical noise, introduced at scales smaller than the skillful scale, e.g. at local scale, was not considered in this study.

中文翻译：

---

比利时中部降水预测的温室气体情景敏感性和不确定性

摘要 由于对未来温室气体浓度和气候模式物理知识的不完整，以及气候系统固有的内部变异性，气候变化对气象变量的影响评估涉及很大的不确定性。鉴于温室气体浓度的变化是这一变化的驱动因素，人们预计其影响将高度依赖于所考虑的温室气体情景 (GHS)。在本研究中，我们将这种行为称为 GHS 敏感性。由于气候模型相关的不确定性，这种敏感性在局部范围内并不总是如预期的那么强烈。本文旨在研究 GHS 敏感性及其对比利时案例研究的气候情景的贡献。考虑了 160 个 CMIP5 气候模型运行的集合，并研究了气候变化信号的降水积累、日降水强度和湿日频率。这是针对一年中的不同季节和 2011-2040、2031-2060、2051-2081 和 2071-2100 情景期进行的。通过方差分解，将气候变化信号的总方差分离为 GHS 差异和其他模型相关不确定性来源的贡献。发现这些贡献取决于变量和季节。遵循出现时间概念，发现 GHS 不确定性贡献取决于时间范围并随时间增加。对于最明显的时间范围（2071-2100），气候模型的不确定性占最大的不确定性贡献。GHS 差异解释了气候变化信号总差异的 18%。结果进一步指出了气候模式集合设计的重要性，特别是集合规模和气候模式的组合，气候情景正是以此为基础的。在本研究中没有考虑在比熟练尺度更小的尺度上引入的数值噪声，例如在局部尺度上。