One of the most difficult aspects of deciding whether current climatic events reveal evidence of the impact of human activities is that it is hard to get a measure of what constitutes the natural variability of the climate. We know that over the past millennia the climate has undergone major changes without any significant human intervention. We also know that the global climate system is immensely complicated and that everything is in some way connected, and so the system is capable of fluctuating in unexpected ways. We need therefore to know how much the climate can vary of its own accord in order to interpret with confidence the extent to which recent changes are natural as opposed to being the result of human activities.

Instrumental records do not go back far enough to provide us with reliable measurements of global climatic variability on timescales longer than a century. What we do know is that as we include longer time intervals, the record shows increasing evidence of slow swings in climate between different regimes. To build up a better picture of fluctuations appreciably further back in time requires us to use proxy records.

Over long periods of time, substances whose physical and chemical properties change with the ambient climate at the time can be deposited in a systematic way to provide a continuous record of changes in those properties overtime, sometimes for hundreds or thousands of years. Generally, the layering occurs on an annual basis, hence the observed changes in the records can be dated. Information on temperature, rainfall, and other aspects of the climate that can be inferred from the systematic changes in properties is usually referred to as proxy data. Proxy temperature records have been reconstructed from ice core drilled out of the central Greenland ice cap, calcite shells embedded in layered lake sediments in Western Europe, ocean floor sediment cores from the tropical Atlantic Ocean, ice cores from Peruvian glaciers, and ice cores from eastern Antarctica. While these records provide broadly consistent indications that temperature variations can occur on a global scale, there are nonetheless some intriguing differences, which suggest that the pattern of temperature variations in regional climates can also differ significantly from each other.

What the proxy records make abundantly clear is that there have been significant natural changes in the climate over timescales longer than a few thousand years. Equally striking, however, is the relative stability of the climate in the past 10,000 years (the Holocene period).

To the extent that the coverage of the global climate from these records can provide a measure of its true variability, it should at least indicate how all the natural causes of climate change have combined. These include the chaotic fluctuations of the atmosphere, the slower but equally erratic behavior of the oceans, changes in the land surfaces, and the extent of ice and snow. Also included will be any variations that have arisen from volcanic activity, solar activity, and, possibly, human activities.

One way to estimate how all the various processes leading to climate variability will combine is by using computer models of the global climate. They can do only so much to represent the full complexity of the global climate and hence may give only limited information about natural variability. Studies suggest that to date the variability in computer simulations is considerably smaller than in data obtained from the proxy records.

In addition to the internal variability of the global climate system itself, there is the added factor of external influences, such as volcanoes and solar activity. There is a growing body of opinion that both these physical variations have a measurable impact on the climate. Thus we need to be able to include these in our deliberations. Some current analyses conclude that volcanoes and solar activity explain quite a considerable amount of the observed variability in the period from the seventeenth to the early twentieth centuries, but that they cannot be invoked to explain the rapid warming in recent decades.

要证实当前气候是否受到人类的影响，最困难的一方面是很难得到一种构成气候自然变化的方法。我们理解到在过去的几千年里，气候在没有显著的人类干预的情况下也经历了重大变化。我们还知道全球气候系统是非常复杂的，各个因素之间存在着千丝万缕的联系，因此，这一系统纷繁复杂变化多端。因此，我们需要知道气候变化在多大程度上是自然原因造成的，从而可以确切地解释出最近的气候变化是自然原因造成，，或相反人类活动的结果。

仪器记录不能追溯到那么久远的时期给我们提供跨度长于一个世纪的全球气候变化的可靠测量方法。我们所确知的就是，当我们想时间跨度延长，记载的信息展现了朝代更迭的过程中气候缓慢的证据。为了更好地描述，更为久远的时期内的气候变化状况，我们需要使用替代性指标。

经过很长一段时间，有些物质地物理和化学特征会随着当时周围的气候变化而变化，它们将会以系统的方法沉淀，这些特征就可以记录更长时间跨度内连续的气候变化,有时可达几百年或几千年。通常来说，分层堆积是每年都会发生，因此我们可以通过记录中可观察的变化来确定时间。温度，降水和气候的其他方面的气候信息通常被称为替代性数据，这些信息可以从这些特征的系统变化中推断出来。替代性温度指标已被重建：格陵兰冰帽中部钻取的冰核，西欧深嵌在分层湖底沉积物中的方解石壳，取自热带大西洋的海底沉积物核，取自秘鲁冰河的冰核,以及取自东南极洲的冰核。尽管这新记录提供了广范一致的迹象指出温度变化可在全球范围内发生，但仍存在有趣的差异，这些差异表明区域性气候的温度变化方式可以如此不同。

替代性指标充分说明，几千年以上的时间跨度里存在着显著的自然气候变化。然而，同样令人惊讶的是在过去的一万年（全新世）中气候变化的相对来说很稳定。

这些指标中记录的全球气候的覆盖范围已经达到了可以提供真实的气候变化标准的程度，它至少能揭示所有引起气候变化的自然原因是怎样息息相关的。这些原因包括变化多端的大气波动，较慢却同样不稳定的海洋活动，地表变化以及冰雪的覆盖度。还包括所有火山活动、太阳活动，可能还有人类活动会引起的变化。

要判断所有导致气候变化的不同过程是如何结合的一种方法就是使用全球气候的计算机模型。它们只可以用来描绘全球气候的全面复杂性，因此只能提供自然变化的有限信息。研究表明迄今为止计算机模拟的气候可变性比从自替代性记录中获取的数据要少得多。

除全球气候系统本身的内部变化之外，还存在其他外部的影响因素，如火山和太阳活动。有越来越多的观点认为这两种物理变化对气候有着可测量的影响。因此我们需要能够考虑到这些因素。一些现有的分析断定火山和太阳活动解释了自17世纪到20世纪早期的相当多的可观察到的变化，但它们不能用以解释最近几十年的全球气候迅速变暖的现象。

替代性指标（substitutive indicator、proxy record）是恢复古气候与古环境的代用信息指标。有关古气候与古环境的信息常被保存在沉积体中，用这些沉积的物质记录可了解较长时间的环境变化。
全新世在地质年表上第四纪后两世从更新世结束一直到现在岩石时期的泥沙时期