Past Climates: How We Get Our Data
On the work floor, research on past climates is known as paleoclimatology, and research on past oceans as paleoceanography. But they are very tightly related, and we shall discuss both combined under the one term of paleoclimatology. Within paleoclimatology, interests are spread over three fundamental fields. The first field is concerned with dating ancient evidence and is referred to as chronological studies. These studies are essential because all records of past climate change need to be dated as accurately as possible to ensure that we know when the studied climate changes occurred, how fast they were, and whether changes seen in various components of the climate system happened at the same time or at different times. The second field concerns observational studies, where the observations can be of different types. Some are direct measurements; for example, sunspot counts or temperature records. Some are historical, written accounts of anecdotal evidence, such as reports on the frequency of frozen rivers, floods, or droughts. Such records are very local and often subjective, so they are usually no good as primary evidence. But they can offer great support and validation to reconstructions from other tools. Besides direct and anecdotal data, we encounter the dominant type of evidence used in the discipline. These are the so- called proxy data, or proxies. Proxies are indirect measures that approximate (hence the name proxy) changes in important climate- system variables, such as temperature, CO2 concentrations, nutrient concentrations, and so on. This chapter outlines some of the most important proxies. The third field in paleoclimatology concerns modeling. It employs numerical models for climate system simulation and simpler classes of so-called box- models. Numerical climate models range from Earth System models that are relatively crude and can therefore be set to run simulations of many thousands of years, to very complex and refined coupled models that are computationally very greedy and thus give simulations of great detail but only over short intervals of time. Box- models are much simpler and faster to run, and they are most used in modeling of the carbon cycle or other geochemical properties.