Proxy data

Proxy data reveal climatic data through the interpretation of some form of related data. For most of the techniques described in this section the essence of that interpretation is quite straightforward. In practice, however, interpreting proxy data requires complex techniques allowing the data to be standardized, that is removing extraneous variations in the data record, and calibrated, thus relating the data to some baseline. The following is therefore a simplified account of the techniques used.


Under this heading is included written data that are not based on instrumental readings but other more anecdotal information. This type of information can be found in hundreds of different written texts. It may be direct information on the weather as in diaries and journals, or indirect evidence, such as crop yields, harvest dates and grain record prices, which will reflect the weather of the time. The data have to be turned from these vague descriptions into some form of numerical description of the weather, so they are a type of proxy data. Obviously there is a great deal of uncertainty in this process which decreases the reliability of the data. For example in the case of grain yield good weather will lead to higher yields and so to lower prices. However, what do we mean by good weather and how would changes in agricultural practice at the time affect yields and prices? What about other social and economic developments that may affect prices? Another problem with such subjective writings is the motive of writer. It may be a diarist with no other motive than to provide an account of the daily goings on in which the weather is a mere adjunct. Or the writer could be making a report as to why an expedition to find new riches failed, in which case the exaggerated weather conditions just might keep the writers head intact, literally. Where possible it is best to compare documentation with other reports and evidence. There is a large amount of material out there, but to pull that information together is very time-consuming. Calibration is usually achieved by attempting to overlap part of the historical documentation with early instrumental data. Some sort of mathematical relationship can then be derived to link the two data sources and thus provide a way of interpreting earlier documentation.

Of course the date will usually appear on the documentation but accurate dating can be a greater problem than you may think. Different countries started their new year on different dates. Then there was the change from the Julian calendar, with no fixed new year's day, to the Gregorian calendar which begins, as now, on 1 January. Apart from the change to the start of the year other alterations to the occurrence of leap years were made. These revisions to the calendar by Pope Gregory XIII happened in most of Europe in 1582 but the Russians changed as late as 1917 after the Bolshevik revolution. Rather than stating the year, some documents refer to being in the X year of the reign of monarch Y. Very often the same weather event is attributed to different years, but sometimes extreme weather can occur in a series of years and it is a mistake to lump them together as one event. So, obtaining the correct year requires careful consideration. Probably the best hoped for time resolution is days, though in ships' logs this might be greater. Annals and account books may describe the monthly or seasonal weather, for example if it was a particularly wet or dry July. This type of data requires that the civilization had developed writing as a means of recording events. Extreme weather events were recorded by the ancient Babylonians on inscribed stones. However, most material comes from Europe, China and India. The information usually refers to the local climate and gives little insight into the global climate of the time.


Ice and ocean cores which have already been discussed in Chapter 3 are also of use in studying historical climate changes. Similarly, varves can be used for historical climate reconstruction as they provide information on rainfall and streamflow. Rivers also provide proxy data through flood records. The variables provided are rainfall and evaporation from flood and lake levels. The length of the time series for some of these records are impressive. Lake levels also provide information on the climate. For Lake Saki in the Crimea the thickness of the yearly mud levels is dependent on the summer rainfall. The length of the climatic record is long and can be detailed back to 2300 bc. Longer time series exist for beds and lakes formed around the margins of ice sheets during post-glacial melting. This has been achieved by using shortlived but overlapping lakes to form one long composite time series. However, the technique is not without difficulties. For the Nile, some flood records go back to 3100 bc, although they mostly date from 622 ad. Flooding in the Nile delta depends on summer monsoons over Ethiopia. The data are really only useful for information about an individual year not a shorter time period. With varves, days or weeks can be ascertained but for lake levels the resolution may be as large as 15 years. Also the data are scattered around the globe from individual lakes and rivers. Reliability of the data also poses a problem. In the Nile prolonged silting changes the zero level making it difficult to ascertain the flood levels.


There are several important ways in which flora and fauna can help with the climatic record. The most well known are the analysis of pollen grains and tree ring analysis called dendrochronology. Pollen grains allow both temperature and rainfall to be assessed and have various features that make them ideal for providing proxy data. Plants produce pollen in profusion and they are extremely resistant to decay. It is possible to identify a plant species from its pollen grain and this will provide information about the climatic conditions of the area. As each pollen grain can be counted, the amount and variety of each species in the locale can be estimated. This rests on the assumption that the pollen grain sample is representative of the plants growing in that area. In some locations, such as peat bogs, the pollen grains are laid down in layers with the peat. Thus changes in the climate can be reconstructed from recording the changes in vegetation. Finally pollen grains can be dated exactly by radiocarbon dating.

Although there is a potential to cover the globe with this type of analysis, in practice it is limited to only a few sites. These sites provide local coverage on a spatial scale but several of the sites provide data stretching back 125 000 years or more. The record length is dependent on site conditions and dating. Unless the pollen is in peat or lake bed sediments it can be difficult to date within 100 years. Furthermore radiocarbon dating is expensive and is limited to 70000 years. The first step in reconstructing a past climate from pollen grains is to determine a species present-day response to climatic indicators, such as temperature and precipitation. This present-day relationship is then used to interpret the past fossil pollen record in terms of the past climate. Problems in reliability lie with the assumption that each forest type corresponds to the same sort of climate as today. Although there is reason to believe that vegetation is fairly responsive to climate, there are known to be some lags. In Europe there are long delays between the establishment of a climate regime and certain trees arriving. It can take centuries for these trees to become dominant. These delays have been detected by the change in insect population which has a quicker response time, particularly some beetle species. In contrast in North America there is little time lag between the climate change and tree response. Human activities have also affected plant species and bog development which again will alter the pollen record.

Dendrochronology allows the determination of temperature and rainfall by looking at the widths of tree rings which have been found to vary in response to changes in rainfall and temperature. Tree rings are formed by many trees. There is a growth spurt in spring when thin-walled cells are formed. The growth then slows down in autumn when thick walled cells are produced. The difference between the two types of cells produces the distinct boundary: the ring. Age is determined by counting the number of rings from the trunk of a felled tree. If the tree was still growing when felled then in the ring widths you have a climatic record from the day it fell, back to when it first grew. In the case of the long-lived Bristlecombe pine this can allow the climate record to go back to 343 bc. The record length that can be obtained, however, is not limited by the tree's life. Although there are small variations in tree ring width, due to local factors, over quite a large area a similar pattern in the rings emerges which is attributable to the climate of that region. The distinctive pattern of the rings then allows that record to be overlapped with those from trees felled earlier, and used for building timber or that have been fossilized or preserved in peat bogs, forming one long record. The determination of the climate from ring widths depends on the tree's response to climate. This is not as straightforward as the description above portrays and depends on the tree biology as well as the climate. Various factors due to the genetic make-up of the tree and the environment introduce variations into the ring widths. Complex techniques are used to remove these variations revealing the trend of the underlying climate variable. Generally, ring widths in trees at the upper altitude limit and at the poleward limit are related to summer temperature. Tree ring width from trees at the warm arid limit are determined more by precipitation. The ring width depends on the previous 15 months weather and so the record has a resolution of about one year. Tree rings are particularly good at showing year to year and decadal variations in the climate (Pearce, 1996).

0 0

Post a comment