Over the past 25 years research concerning the effects of climatic fluctuations on past human societies has shifted focus considerably, with most recent hypotheses emphasizing shorter-term variability over longer-term change. Definitions of change and variability, however, remain subjective and vary considerably between researchers. It is suggested that white noise, due to its inherent unpredictability, provides a theoretically robust model of variability that accords with perceptions of variability conveyed by the existing literature. The use of white noise as a model for variability enables the development of an algorithm that objectively decomposes an empirical climatic signal into change and variability components. The algorithm, which combines singular spectrum analysis and Fourier methods, is validated via an extensive series of simulations and applied via two empirical case studies. It is shown that the algorithm has the potential to produce genuine advances by isolating features of interest and facilitating more rigorous hypothesis testing. Its use will therefore aid researchers studying palaeoclimatic effects on prehistoric human societies as well as those studying the nature and effects of contemporary climate change.

在过去的 25 年中，关于气候波动对过去人类社会影响的研究已经发生了很大的变化，最近的假设强调了短期变化而非长期变化。然而，变化和可变性的定义仍然是主观的，研究人员之间差异很大。有人认为，白噪声由于其固有的不可预测性，提供了一个理论上稳健的变异模型，该模型与现有文献所传达的变异感知一致。使用白噪声作为可变性模型可以开发一种算法，将经验气候信号客观地分解为变化和可变性分量。该算法结合了奇异谱分析和傅立叶方法，通过一系列广泛的模拟得到验证，并通过两个实证案例研究加以应用。结果表明，该算法有可能通过隔离感兴趣的特征并促进更严格的假设检验来产生真正的进步。因此，它的使用将有助于研究古气候对史前人类社会的影响以及研究当代气候变化的性质和影响的研究人员。