Scales and Scaling in the Climate System

传统上，气候被视为在相对较窄的时间范围内各自作用的过程的复合体。最明显的是与每日和年度周期有关的周期，但还有许多其他周期：例如，厄尔尼诺和与冰期有关的轨道(米兰科维奇)周期。“介于两者之间”的音阶是乏味的“背景噪音”。然而，大多数可变性是在这些大范围的(“缩放”)过程中：它们是前台过程，而不是后台过程。这一提议的核心是重新确立这些缩放过程的首要地位，并研究了解过去、现在和未来气候的后果。*理解过去的气候在很大程度上是一个解释从冰和海洋核心到树木年轮和湖泊沉积物的古“替代”气候记录的问题。这项提议将使用新的“尺度涨落分析”技术来系统地理解和模拟从数百年到数百万年的时间尺度，从行星大小到数百公里的空间尺度的变异性。这是重建过去所必需的。*对于目前的气候，该提议利用“蝴蝶效应”将天气降低为随机噪音，驱动具有长期记忆的系统。蝴蝶效应指的是大气非常不稳定，以至于一只蝴蝶拍打翅膀就可以改变它的航线。虽然“蝴蝶效应”限制了天气预报的准确性，但在十天以外的几个月甚至更长的时间里，天气几乎没有什么技能，天气有效地给大气状态提供了随机的“轻推”，而轻推的统计数据与巨大的系统内存相结合，可以用来做出长期预报。事实证明，大气的记忆力如此之强，以至于我们仍能感受到百年波动的影响，这种记忆力可以被利用，从而改善出了名的糟糕的季节性预报。*未来气候不仅取决于与长期记忆和蝴蝶效应相关的自然变化，还取决于大气成分，特别是温室气体-土地利用和其他人为变化。目前，数值气候模型对2050年或2080年的预测存在很大分歧。为了给我们对未来排放情景的信心，需要存在一个新的、性质不同的未来预测来源。*在未来五年，对于过去、现在和未来的气候，我们的目标是a)产生新的多重代理重建，b)产生新的月度、季节、年度和年代际预报，c)产生对21世纪的新的温度和降水预测。通过利用过去的历史数据在没有模型的情况下进行预测，从而绕过模型的不确定性。**

The climate has traditionally been viewed as a composite of processes acting each over relatively narrow ranges of time scales. The most obvious are those associated with the daily and annual cycles, but there are many others: for example El Nino and the orbital (Milankovitch) cycles associated with the ice ages. The “in between” scales are are uninteresting “background noises”. However, most of the variability is in these wide scale range (“scaling”) processes: they are foreground, not background processes. This proposal is centred on re-establishing the primacy of these scaling processes and examining the consequences for understanding the climate of the past, present and future.***Understanding the past climate is largely a question of interpreting the paleo “proxy” climate records from ice and ocean cores, to tree rings and lake sediments. This proposal will use new “scaling fluctuation analysis” techniques to systematically understand and model the variability over time scales from centuries to millions of years, over spatial scales from the size of the planet to hundreds of kilometers. This is needed in order to reconstruct the past.***For the present climate, the proposal exploits the “butterfly effect” to reduce the weather to random noise driving a system with a long term memory. The butterfly effect refers to the fact that the atmosphere is so unstable that a butterfly flapping its wings can change its course. While the "butterfly effect" limits the accuracy of weather forecasts these have little skill beyond ten days at scales of months and longer the weather effectively gives random “nudges” to the state of the atmosphere and the statistics of the nudges combined with the huge system memory can be used to make long range forecasts. It turns out that the atmosphere's memory is so strong that we are still feeling the effects of century old fluctuations, and this memory can be exploited, thus improving the notoriously poor seasonal forecasts. ***The future climate depends not only on the natural variability associated with the long range memory and butterfly effect, it depends on the atmospheric composition especially greenhouse gases - land use and other anthropogenic changes. Currently, numerical climate models disagree significantly with each about the projections to 2050 or 2080. The existence of a new qualitatively different source of future projections is needed in order to give us confidence in future emission scenarios.***In the next five years, for past, present and future climate we aim to a) produce new multiproxy reconstructions, b) produce new monthly, seasonal, annual and decadal forecasts, c) produce new temperature and precipitation projections for the 21st century. bypass the model uncertainties by making projections without the models by exploiting the past historical data. **