Climatology: Methods
Climatology: Methods
- Sharon E. NicholsonSharon E. NicholsonFlorida State University
Summary
Environmental constraints have large impacts on populations, especially in semi-arid regions such as Africa. Climate and weather have long affected African societies, but unfortunately the traditional climatic record for the continent is relatively short. For that reason, historical information has often been used to reconstruct climate of the past. Sources of historical information include reports and diaries of explorers, settlers, and missionaries; government records; reports of scientific expeditions; and historical geographical and meteorological journals. Local oral tradition is also useful. It is reported in the form of historical chronicles compiled centuries later. References to famine and drought, economic conditions, floods, agriculture, weather events, and the season cycle are examples of useful types of information. Some of the records also include meteorological measurements. More recently chemical and biological information, generally derived from lake cores, has been applied to historical climate reconstruction. Early works provided in most cases qualitative, discontinuous information, such as drought chronologies. However, a statistical method of climate reconstruction applied to a vast collection of historical information and meteorological data allowed for the creation of a two-century, semi-quantitative “precipitation” data set. It consists of annual indices related to rainfall since 1800 for ninety regions of the African continent. This data set has served to illustrate several 19th-century periods of anomalous rainfall conditions that affected nearly the entire continent. An example is widespread aridity during several decades early in that century.
Subjects
- Historical Preservation and Cultural Heritage
- Historiography and Methods
Climate and Society
Climate and weather have a tremendous impact on human societies. They dictate agriculture and human comfort and affect the occurrence of many diseases. Vagaries of climate such as droughts and floods create misery and economic hardship, in some cases leading to human migration. A decades-long drought in the 1830s, for example, prompted the exodus of several peoples from Uganda.1 This drought also affected southern Africa, causing a breakdown of social, political, and economic institutions among the Nguni cattle-keeping cultivators, giving rise to Shaka’s Zulu kingdom.2 During the mid-18th century, a two-decade period of drought prompted frequent invasions of the Oyo into Dahomey (current-day Benin) during the most intense years of drought.3 Thus, climate can affect history, and its changes over time are of interest to the historian. This is particularly true in the case of Africa, where semi-arid conditions prevail and much of the population lives off the land.
This brief overview of climatic methods considers three topics: sources of data and information, quantitative climatic analysis, and methods of reconstructing climatic history during the historical period of roughly the last two thousand years. Examples are included of select reconstructions and their relationship to history. A map showing the locations cited in the text is given in Figure 1.
Figure 1. Map indicating locations mentioned in the text (countries, cities, lakes, and rivers). Shading indicates a general area referred to in the text.
Sources of Data and Information
Modern Quantitative Data
Observational Data Sets
The most important data for understanding the past are records of temperature and precipitation. These variables are recorded daily at thousands of stations across Africa, in virtually every African country. However, only a portion of what is recorded eventually ends up in a readily accessible data archive outside of that country. Unfortunately, there are also major gaps in the records because of turmoil within various countries. This is particularly true for Angola and the Democratic Republic of the Congo, where the meteorological services nearly ceased to function for one or more decades.
Precipitation measurement has a longer history than temperature measurement. Nevertheless, few observations are available prior to 1900, except for South Africa, Algeria, and Tunisia. Coverage was fairly dense in South Africa and Algeria as of the mid-19th century and in Tunisia as of about 1885. Other notable long records are shown in Table 1. The number of recording stations dramatically increased as of the 1920s, with peak coverage for Africa as a whole probably being in the 1950s. The patterns of availability relate closely to the colonial history of the various countries. A majority of countries became independent in the 1960s, after which time national agencies replaced the colonial powers and in many cases economic and political upheaval led to a decline in meteorological observations.
Nevertheless, long records from precipitation gauges can be derived for about one and a half centuries or more for several parts of the continent.
Figure 2. Annual rainfall for several Africa regions, expressed in units of standard deviations from the long-term mean. Locations are shown in the inset.
Figure 2 gives examples for the Sahel, with continuous series beginning in 1847, East Africa beginning in 1874, coastal Algeria/Tunisia, beginning in 1838, the Guinea Coast beginning in 1883, semi-arid southern Africa beginning in 1854, and the Cape winter rains region of South Africa beginning in 1838.
In the late 1960s, when attention was focused on Africa as a result of drought in the Sahel, extensive precipitation records were not readily available to researchers. As a result, some of the early assessments4 were based on a very small handful of stations. Since that time much effort has gone into digitizing and making available temperature and precipitation data sets for Africa. The best known are the Global Historical Climatology Network (GHCN) assembled by NOAA, the Global Precipitation Climatology Centre (GPCC) archive that is the responsibility of the German Weather Service, and three data sets compiled by the University of East Anglia, and the University of Delaware. In general, considerably more precipitation data are available than temperature data, because precipitation is recorded at more many sites. However, temperature is much more spatially coherent than is precipitation, so that fewer records are needed to develop a reliable picture of climate.
For the most part, the early precipitation records in each of these archives (up to at least the 1980s and in some cases up to 1998) were assembled by one individual, the author of this chapter. For that reason, the data sets cannot be considered to be “independent,” so that an agreement of records in two data sets does not provide evidence of the quality of the data. Those records were acquired mainly through a compilation of observations published by each individual country and by several sets of writing campaigns, in which monthly rainfall for select stations was requested from the individual weather services of each country. In addition, many of the records for Francophone countries were provided by ASECNA (Agence pour la Securité de la Navigation Aerienne), the French equivalent of the American Federal Aviation Agency. The author also made her African data set through 1998 available via inclusion in GHCN or by special request.
After that time the other archives were updated primarily in three ways: compilation of the daily data transmitted via the Global Telecommunications System (G.T.S.), by cooperative agreement with some countries, or by ad hoc donation from individuals. The purpose of G.T.S. transmission is real-time forecasting. For that reason, if data are not transmitted on time, generally no attempt is made to obtain late reports. Thus, there are big gaps in the G.T.S. records, and some countries are more cooperative than others in providing the data.
Figure 3. Temperature and precipitation stations in the G.T.S. network plus the availability of precipitation data in one individual month (June 1989).
Figure 3 shows the temperature and precipitation stations in the G.T.S. network, as well as an example of availability in one given month (June 1989).
The five major archives differ with respect to which variables are compiled, how the data are processed, and how the data are presented. All include precipitation data for all African countries. The data sets of the University of East Anglia (also termed CRU data, for Climatic Research Unit) and the University of Delaware also include temperature data. Only GHCN and the Nicholson archive provide data for individual stations. The others present gridded (i.e., spatially averaged) values at various resolutions. The reason is twofold. For many climate applications, such as comparison with forecast models, gridded data are appropriate. Moreover, the formal agreements made with most African countries preclude the dissemination of actual station data. These are generally available from individual meteorological services, but in some cases at great cost.
GPCC is precipitation only and the data run from 1901 to the present time, although the most reliable version for climate studies extends only to 2013. African precipitation is available from GPCC at three different resolutions: 0.5, 1.0, and 2.5 degrees.5
The full complement of data available from CRU is described at their website. It includes, in addition to monthly mean air temperature and monthly precipitation, records of maximum and minimum temperature and many other meteorological variables. A land precipitation data set running from 1901 to 1998 is available at both 5 degree latitude/longitude resolution and 2.5 degree by 3.75 degree resolution.6 The University of Delaware data set also runs from 1901 to 2014, but at a resolution of 0.5 degrees. Many water-balance parameters are also provided.7
GHCN includes monthly mean temperature and monthly precipitation totals for individual stations. Although all African countries are included, the coverage depends greatly on the country and also varies over time. The most recent data are generally obtained through the G.T.S. system, and for that reason most recent records are spotty, with many individual months and years missing. Coverage is relatively complete for most countries up through 1998. For most countries, only a select group of stations is included.
Satellite and Blended Data Sets
In order to obtain global climate coverage, satellites and models have been used to generate gridded fields of meteorological variables. For precipitation estimation, several types of satellite methodologies are utilized, but some of the best blend satellite estimates with surface gauge data. In addition, several products termed “Reanalysis” apply meteorological models to a combination of satellite, surface, aircraft, and upper-air data from radiosondes and pilot balloons. Blended data sets combining satellite and gauge information are usually the most accurate for precipitation estimates.8 However, their performance varies considerably in different parts of Africa.
While satellite data are important tools for the climatologist, they are too short in duration to be useful for historical studies. Although the first weather satellites were launched in the 1960s, climate data sets produced via satellite generally extend back only to 1979, at best. Two reanalysis products do cover a much longer period. The Twentieth Century Reanalysis of NOAA extends from 1851 to 2014.9 The ERA-20CM data set of the European Center for Medium Range Forecasting covers 1900 to 2010. However, the performance of these reanalysis products over Africa has never been examined.
Proxy Data Sources—Chemical and Biological
Because the meteorological record for Africa is relatively short, proxy data become very important in reconstructing climate during the historical period. In temperate latitudes, tree rings have been a tremendous source of information. Unfortunately, few tree species in Africa have appropriate annual rings. North Africa is the one region for which tree-ring analysis (dendrochronology) has been very useful in establishing long-term records of rainfall.10 However, a wide variety of other methodologies have been developed and utilized elsewhere.
In West and equatorial Africa, sediment cores from lakes have provided a wealth of information on precipitation. Some of the relevant properties that have been utilized include sedimentation rate and character; carbonate content; salinity; diatoms; pollen and spores; clay weathering; levels of magnesium, potassium, and biogenic silica; deuterium (heavy hydrogen) in leaf wax; and strontium/calcium ratios and oxygen isotope records in ostracod shells.11 Oxygen isotopes in stromatolites (layered sedimentary rock formations of microbial origin), varved sediments, and ancient lake shorelines have also proven useful.12 Varves are layered lake sediments with contrasting characteristics in each couplet of varves. The contrast represents annual cycles in sedimentation or biological and chemical processes, which in turn generally relate to the annual cycle of precipitation or temperature. For that reason, they provide a high-resolution record of the past. African lakes with varves include Bosumtwi, Challa, and Malawi.13 The dust record in marine sediment cores offshore from Senegal also provides climatic information.14
In southern Africa, few records are well enough dated to examine variations on timescales of use for historians. Exceptions are oxygen and carbon isotopes in speleothems (cave deposits such as stalagmites and stalactites), carbon isotopes in baobab trees, pollen, hyrax and shell middens, diatoms, and marine cores off the coast of Namibia.15 The latter provide records of SSTs and upwelling that generally correlate with precipitation. A handful of tree-ring studies have also been carried out in southern Africa, but these generally provide only a century of information.16 The exception is a 413-year chronology from the Cape Province of South Africa.17 Although tree rings in subhumid areas such as southern Africa tend to reflect precipitation, their interpretation is complicated by other factors affecting moisture availability, such as growth patterns, temperature, and orographic effects.
In Africa, where precipitation is generally the limiting factor in environmental processes, there are few sources of proxy information to reconstruct past temperatures.18 Moreover, the interpretation is very complex, much more so than with precipitation. The few available temperature proxies are generally available only for a handful of locations in East and southern Africa, so that there is a void over the rest of the continent. A limited number of pollen and chironomid (insect) records have been interpreted in terms of temperature. Bore-hole records provide an indication of temperature, but at relatively low resolution (i.e., century-scale). Oxygen isotope records from cave speleothems in southern Africa and ice cores in East Africa have also been useful.
A methodology termed TEX86 provides the most continuous and highest-resolution paleotemperature records for Africa. These have been derived for Lakes Turkana, Tanganyika, and Malawi. The method involves analysis of lipids in the microbe Thaumarchaeota.19 The TEX86 records for Tanganyika and Malawi respectively extend back some 1,500 and 700 years. A variant of the method, based on ratios of characteristics of the lipids, shows further potential for paleo-temperature reconstruction on the historical time scale.
Proxy Data Sources—Historical and Geographical
Some of the first historical information on African climate was provided by classic sources such as C. E. P. Brooks’s Climate through the Ages, Brooks and P. A. Buxton’s Le Climat de Sahara et de lArabie, H. H. Lamb’s Climate and History, and Le Roy Ladurie’s Histoire du climat depuis l’an mil.20 Notably, Brooks and Lamb were climatologists, while Le Roy Ladurie is a historian, thus showing the interdisciplinary interest in this field. Each of these sources contained examples of African climate change gleaned from historical, archaeologic, and geographical sources. What was probably the first modern source to focus on Africa, that of Pejml, was written in Czech. Pejml, an agricultural meteorologist, gathered historical climate information as a hobby.21 He demonstrated the usefulness of travel journals in such reconstructions, paving the way for many later studies. Plote, a hydrologist, similarly compiled a combination of documentary and quantitative information on historical climate change in West Africa.22
A wealth of historical and geographical information for earlier centuries exists for Africa and has provided the most widespread and detailed reconstructions of African climate over the last two to three centuries. The principal types include reports and diaries of explorers, settlers, and missionaries; government records; reports of scientific expeditions; and local oral tradition, generally reported in the form of historical chronicles documenting the reigns of centuries of kings. Most information covers brief periods of time and multiple locations. A notable exception is missionary records, which often relate to a single location and cover a longer period of years. Ships’ logs have seen some application and some hydrologic records, notably that of the Nile, also extend to earlier centuries.
Information from Explorers, Settlers, and Missionaries
Africa in the 19th century was a hub of exploration and scientific study for many European countries. Major powers—the French, Italians, Germans, Belgians, British, Danish, and Portuguese—established colonial governments throughout the continent. Exploration and expeditions were a form of entertainment for the folks back home, and extensive reports were published as books and in geographical journals.23 The most important of the latter for meteorological data were Petermann’s Geographische Mitteilungen and Meteorologische Zeitschrift, as well as geographical journals of several national societies. Explorers such as Barth generally included weather diaries in their narratives.24 A few actual quantitative meteorological observations were published. An example is monthly rainfall measurements at Christiansborg, Ghana, for ten years in the 1820s, 1830s, and 1840s. Multi-year compilations were also published by numerous authors, such as Borius, Raulin, and Rousseau.25 More common were reports of famine, drought, the conditions of crops, unusual weather, the course of the seasons, and the status of vegetation, lakes, and waterways.26 Additional information on sources is provided in Nicholson.27
This type of information is particularly useful for the 19th century. However, the record goes back considerably further for some parts of Africa. Information for the Guinea Coast and Angola is plentiful throughout the 18th century as well. A number of European explorers traversed West Africa during that time, and most provided meteorological observations during their journeys. The Cape region of South Africa was settled by Europeans considerably earlier, so that some observations extend back to the late 17th century.
Oral Tradition
Africa has a rich historical legacy, but much is in the form of oral tradition. These traditions are often written down after the fact, sometimes several centuries later. The best known from West Africa include those of Tichitt, Oualata and Nema, the Wâlo Empire, Timbuktu, and the Bornu Empire.28 Each of these contains references to major periods of drought or famine and periods of prosperity. Although the timing of events is unfortunately not precisely dated, there is considerable agreement among chronicles overlapping in time. Thus, one can have a fair degree of confidence in the approximate timing of events going back several centuries. Oral tradition also gives useful information on such factors as migration, disease, agriculture, and landscape (e.g., conditions of lakes and rivers). These factors do have a relationship to climate and can contribute to climate reconstruction, especially when used in conjunction with other information.
Two studies have used oral tradition to reconstruct environmental conditions in the upper Nile region of East Africa.29 These were partially dated based on written records of Nile flow, with assumptions about the length of various rulers. The dating is probably less precise than that from the West African chronicles.
In many cases, oral traditions have been passed down to and recorded by European explorers. These similarly provide useful information for climate reconstruction. One example comes from the Lake Ngami region of Botswana. In the mid-19th century, it was reported to be a vast body of water with waves strong enough to push hippos to shore. Paintings from this period document its expanse (Figure 4).
Figure 4. Lake Ngami sketch. David Livingstone, Missionary Travels and Researches in South Africa, Including a Sketch of Sixteen Years’ Residence in the Interior of Africa, and a Journey from the Cape of Good Hope to Loanda on the West Coast, Thence Across the Continent, Down the River Zambesi, to the Eastern Ocean. London: John Murray, 1857.
Around 1820 a very old man claimed that during his youth Lake Ngami had completely dried up, replaced by a tree-lined stream running through a grass plain. Originally little credence was given to this story. However, when Lake Ngami later dried out in more modern times, the remnants of the streamed and trees became visible (Figure 5).
Figure 5. A dry riverbed in the middle of a grassy plain that was formerly covered by Lake Ngami (Botswana).
Hydrologic Information
Many sources include information about the state of lakes and rivers. In some cases, the information is precise enough that wet and dry periods of the past can be identified. Lakes Malawi and Chad serve as two examples, with historical information plentiful enough that trends over several centuries could be reconstructed. In the case of Malawi, archaeological data also helped to reconstruct its general trends since the 3rd century ce.30 Historical information has allowed for the reconstruction of the levels of numerous other lakes and river in eastern and southern Africa during the last two to three centuries.31
An example of a useful oral tradition is a reference to a dry period of the lake reported by Johnson.32 He was told by a local that the north end of the lake had been so dry seventy years earlier (around the 1820s) that a chief and his men could wade across the lake in areas later “fathoms deep.” Some confirmation comes from a story about Ngonde king Mwangonde, whose reign was genealogically dated to about 1815 to 1835. He waded across the northern part of the lake to marry a woman.33 Such references are plentiful in African history.
In the case of Lake Chad, historical information was used by Maley and Nicholson to reconstruct its levels over several centuries.34 One example is an old man reporting to the explorer Heinrich Barth in 1851 that in his youth one could travel by canoe from Lake Chad to the north of the country.
Figure 6. Photos of Lake Chad taken from space in four different years. The 1963 photo was taken by an astronaut, the others from satellites.
This would be tens of miles along a now dry waterway. This story at first glance sounds somewhat incredible, but evidence from additional sources confirmed that very wet conditions prevailed in the Sahel toward the end of the 18th century. Lake Chad and similar shallow lakes are particularly useful in historical reconstructions because their spatial extent is highly sensitive to precipitation fluctuations. Such changes in recent times are readily seen from photos of Lake Chad taken from space (Figure 6).
The most continuous and longest hydrological record for Africa is that of the Nile flow. Its level was recorded at temples along its banks even several millennia ago. A quantitative record from the Rodah nilometer, in Cairo, commenced in the year 622 ce. Monthly maximum and minimum levels were recorded, along with several statistical analyses.35 This record is without a doubt useful, but caution must be taken in its interpretation. It jointly reflects the flows of both the White Nile, originating from the outflow of Lake Victoria, and the Blue Nile, coming off the Ethiopian highlands. Although it is often assumed that its summer minimum reflects the former and its flood stage the latter, the interpretation is much more complex. Moreover, the Nile levels measured at the Rodah gauge have been influenced by other factors from siltation to possible intentional alteration as a means to justify tax levies. Another important hydrological record is that assembled by Sieger, who published a useful historical chronology of numerous African lakes.36
Climatological Methods
The Character of Climate Change
Climatic change can take several forms, as illustrated in Figure 7. One is periodic variation, which is values of an element repeating themselves at regular intervals. No climatic element is strictly periodic, but may exhibit changes that tend to recur at select but irregular intervals. This is termed a quasi-periodicity.
Figure 7. Forms of climatic variability and change.
More common are discontinuities or jumps (abrupt changes) or upward or downward trends. The Sahel time series in Figure 2 well illustrates a discontinuity, with the onset of drought in 1968. Climatic elements may also exhibit a change in variability over time. This is readily seen in the rainfall time series for the “short rains” (October–November) of East Africa. Rainfall was almost continuously below normal from 1920 or earlier to 1960, and year-to-year changes were relatively small. An abrupt change to wetter conditions occurred in 1961, after which time the year-to-year changes markedly increased as well. Also to be considered in terms of rainfall is a change in the seasonal cycle. This is illustrated for the station Aburi, along the Guinea Coast of Ghana. While in the mean there is a major rainy season in the boreal spring and a minor peak in the boreal fall, in some years the normally dry boreal summer is extraordinarily wet.
These variations have different effects on a population. A trend gives a population time to adjust to changing conditions. An abrupt shift can be catastrophic, as was the drought that commenced in the Sahel in 1968. Although the 1980s were worse, the consequences were less catastrophic as some societal adjustments (e.g., urbanization and mortality) had occurred. Increased variability is also difficult because it increases the probability of extreme events, such as droughts and floods. In East Africa the extreme variability of the 2010s was disastrous. Floods occurred in early 2010 and late 2011, but there was tremendous drought during the intervening months (Figure 8).
Figure 8. Time series of monthly rainfall in equatorial East Africa, expressed as a percent departure from the long-term mean.
A change in the seasonal cycle can be similarly calamitous, as activities such as agriculture are adjusted to the typical seasons. There can be year-to-year changes in the seasons, as illustrated with the station Aburi (Figure 9).
Figure 9. Seasonal cycle at Aburi, Ghana, in mm per month, for three contrasting years.
However, longer-term changes can also occur. This is illustrated with the station Casablanca in Morocco (Figure 10).
Figure 10. Seasonal cycle in Casablanca, Morocco, five-year averages of monthly rainfall, plotted every five years.
Earlier in the 20th century, the rainy season tended to be bimodal (i.e., two peaks in the annual cycle, in the boreal spring and autumn), but toward mid-century it shifted to unimodal with a single peak in the boreal winter.
Quantitative Methods of Analysis
The typical questions asked by the climatologist are how conditions change over time, what is the spatial pattern of climate, and how one location connects with another or with factors regulating change. These questions are also of interest to the historian.
Various statistical and analysis methods are used to assess the temporal and spatial variations of climate and the spatial interrelationships. The most common methods of evaluating temporal change include trends, spectral analysis, and wavelet analysis. An analysis of trends determines the overall tendency of a time series over a select period of time. The result of the method is a straight line calculated so that the square of the departures from the line are minimized. Unfortunately, the result depends very much on the time period considered. Because of this dependence on period considered, a pure trend analysis is not necessarily useful. Its greatest application is in determining how many locations show a particular pattern or how one location or variable relates to another.
Two other methods of temporal analysis ask the question as to whether the time series tends to vary on a particular time scale. That is, is the time series periodic or quasi-periodic? For example, what is the typical interval between droughts? One approach is termed spectral analysis and is based on the assumption that the character is stationary in time. A second approach, termed wavelet analysis, shows how the spectral character changes over time. The details of the calculations are beyond the scope of this chapter, but examples are given in Figures 11 and 12.
Figure 11. Spectrum of East African precipitation during the period 1901 to 1973.
The spectrum of East African rainfall shows three significant peaks at roughly 5.6, 3.6, and 2.2 years during the period 1901 too 1973.37 These same peaks are evident in el Niño, suggesting a link with East African rainfall.
A wavelet analysis of the East African “short rains” is shown in Figure 12. The interpretation is much more complex.
Figure 12. Wavelet analysis of October–November rainfall over East Africa and winds over the central equatorial Indian Ocean.
The y-axis is the time scale in years, and the x-axis is a nondimensional index of intensity of the peak. Although peaks on these same time scales are apparent, the importance of each peak varies in time. Since roughly 1990, variability has been concentrated on scales of roughly two to three and five years. This is actually evident in the East African time series in Figure 2. The interval between the post-1990 wet years is generally two years, with a five-year gap between clusters of years. In contrast, in intervals surrounding the 1960s, when a dramatic change occurred from dry to wet conditions, variance is concentrated on several long and short time scales.
Spectral and wavelet analysis are both useful in comparing the relationship between two variables, although they do not show any definitive links. A simpler method of comparing two variables is linear correlation. In this case, the departures from the mean of each variable at each point in time are compared, and their squares are tallied. The value of the correlation coefficient r ranges from −1 to +1. The latter indicates a perfect correlation: the value of one time series is an exact multiple of the other at each point in time. The former indicates the same, but with the sign of one variable being opposite the sign of the other. The square of the correlation coefficient gives the percent of common variance in the two series. That is, if r is .8, one time series can predict 64 percent of the variability of the other.
Figure 13 illustrates negatively and positively correlated rainfall series. The top diagram shows annual rainfall (expressed in units of standard deviations from the mean) for areas of southern Africa that are roughly 1000 km apart.
Figure 13. Top: time series of annual rainfall for two regions of southern Africa that are roughly 1000 km apart. The correlation between them is .72. Bottom: rainfall series for areas within Malawi and South; the correlation between them is −.25. Location of regions (indicated by numbers in upper left) can be seen in Figure 21.
The similarity is obvious, and the correlation between them is +.72. The bottom diagram compares a region in East Africa with a region in southern Africa. The correlation between them is −.25, much smaller but still statistically significant. Linear correlation is an extremely common climatological method but it has its shortcomings. For one, it does not indicate cause and effect. For another, the value is strongly affected by extreme values. Finally, correlations can arise by chance, so that testing the significance of a correlation coefficient is of paramount importance.
Also important is the depiction of spatial patterns of climatic elements. Of interest is both the mean pattern and patterns of departure from the mean (anomalies) during specific events. Over most of Africa, the patterns of temperature are relatively uniform, and temperature is not extremely variable from year to year, except in the extreme temperate margins. The seasonal cycle is often affected by the clouds of the rainy season, rather than the annual cycle of radiation that is so important in higher latitudes. Thus, in Sahelian West Africa for example, the hottest months are April and May, just prior to the rainy season. Precipitation is much more variable in space (as well as in time) because of the strong influences of coastal currents and topography. The latter factor is well demonstrated by a map of mean rainfall over Uganda (Figure 14).
Figure 14. Mean annual rainfall over Uganda, in inches.
A frequent method of examining climatic change or weather events such as drought is composite analysis. The spatial pattern of elements is compared with the mean conditions or over two periods of time, for example a wet year and a dry year.
Figure 15. 1950s rainfall anomalies averaged over a 1 degree latitude/longitude grid and expressed as a percent departure from the long-term mean.
Figure 15 shows rainfall averaged (i.e., composited) for the 1950s and depicted as a percent departure from the long-term mean. Most of the continent, except for the equatorial regions, was relatively wet.
Figure 16. Rainfall in 1958 minus rainfall in 1984 (in mm), otherwise as in Figure 15.
Figure 16 shows the difference between a very wet year (1958) and a very dry year (1984). Most of the continent was notably drier in 1984.
Climatologists frequently use a method termed Empirical Orthogonal Functions (EOFs) or Principal Component (PC) Analysis to quantify the most recurrent spatial patterns and their frequency of occurrence over time. Figure 17 shows an example of the first four EOFs for rainfall variability over North Africa from gauge data for the 20th century (1920–1994) and for a historical rainfall reconstruction for the 19th century (1820–1900).
Figure 17. The four most important EOFs of Sahel rainfall, based on 1920 to 1994 (left) and based on 1800 to 1900 (right). Colors represent the degree to which each location is represented by the EOF. For example the first EOF indicates a pattern of above-average rainfall throughout the region, but most strongly developed over the Sahel. The second shows above-average rainfall over the Sahel but strongly below average over the Guinea Coast. The sign convention in such an analysis is arbitrary, so that the EOFs basically show (1) in-phase variations throughout the region and (2) an out-of-phase relationship between the Sahel and Guinea Coast.
The first (i.e., most common) pattern shows coherent variability through the region. Note that the sign of the determined EOF is arbitrary, so that the EOF pattern is both what is shown and inverse of the pattern. The next three EOFs also show an opposition between the Sahel and Guinea Coast or the equatorial region in general. The close agreement between the modern and historical patterns gives confidence to the results and strongly suggests that these patterns are inherent modes of variability over North Africa that probably held throughout at least historical times. Alternate methods of pattern identification are described by Lund, including a variant of linear correlation applied by Blasing and Nicholson.38
A common task in climatology is regionalization of data. The essence is to define regions within which a particular climatic variable behaves fairly coherently over time. The importance of this is to minimize the error inherent in looking at individual locations, especially for variables like precipitation. PC analysis can be applied to this. More often statistical methods referring to a “clustering” are instead applied.39 The optimal approach may be a combination of PC analysis and clustering.40
Climatologists often use more complex statistical methods, but these are usually variants of the basic methods already described. Some of these methods are very useful in reconstructing historical climates. They allow for the determination of fields of meteorological elements from data sets that are spatially incomplete and variable over time. Typical methods of reconstruction are described by M. E. Mann and P. D. Jones, R. Neukom and colleagues, and J. E. Smerdon and colleagues.41
Historical Climate Reconstruction
The modern quantitative methods might be of limited use to the historian because regular observations generally do not extend back beyond the early or mid-19th century. Several historical reconstructions for earlier centuries have been produced by various authors. Nicholson and Brooks developed a climate chronology for the Sahel covering one to two millennia, with emphasis on the last few centuries.42 The former was based on a wide variety of historical and geographical sources relating environmental conditions, while the latter reconstruction was based primarily on human reaction to rainfall variability. Nicholson later extended the analysis to the entire continent.43 While a very generalized historical time series was published for the Sahel, the bulk of the reconstruction consisted of identifying periods of anomalous rainfall conditions. The work eventually evolved into a semi-quantitative data set extending from 1800 for ninety regions covering the entire continent of Africa.44
Several other authors have published more regionally focused reconstructions. Nash and co-authors published a number papers dealing with rainfall variability, as well as the occurrence of snow and of tropical cyclones, in several parts of southern Africa during the 19th century.45 Hannaford et al. reconstructed rainfall in southern Africa at the seasonal scale using wind observations in ships logs.46 Anderson used documentary evidence to chronicle hydroclimate variability in western Kenya since 1750.47 Norrgård published a rainfall chronology for the Guinea Coast covering the period 1750 to 1800.48 Each of these reconstructions provided semi-quantitative information in the form of an index ranging from −3 to +3 (from extremely dry to extremely wet) or from −2 to +2. Others have established historical drought chronologies that are principally descriptive. Nicholson published multi-century drought chronologies for Algeria, Senegambia, Chad, the Niger Bend, and several locations in southern Africa.49 Webster and Herring derived chronologies for equatorial Africa, using Nile records to assist in dating drought references from oral tradition.50 Historical drought chronologies have also been published for Ethiopia and Angola.51
In what was probably the first paper to describe a distinct methodology for historical climate reconstruction over Africa, Nicholson described two approaches.52 One is to determine a long-term trend, that is, to determine if periods in the past were wetter or drier, colder or warmer, than today. Unfortunately, this requires the compilation of nearly continuous series of a climatic element. Such information is rare, but in some locations enough historical detail is available that reliable drought chronologies can be produced. Figure 18 shows some examples from widespread regions of Africa.
Figure 18. Multi-century chronologies of drought and wetter conditions in select African regions.
These have been compiled from a large number of sources, including those mentioned in the previous paragraph plus Abdelhadi, Almeida, Becker, Brooks, Marchika, Nicholson, and Patterson.53
The second approach determines short-term climatic anomalies that occurred in the past, for example, widespread reports of drought or dry conditions clustered in time. Figure 19 shows an example for the period 1828 to 1839, based solely on historical evidence. The reports suggested a continent-wide period of drought.
Figure 19. Precipitation anomaly map for the period 1828 to 1839, based on information in Nicholson (1976, 1979). Historical information is available from each indicated location, with squares indicating lakes and rivers and plus and minus signs indicating abnormally wet or dry conditions, respectively. The numbers indicate the number of years, within the period 1828–1839 in which drought conditions were referenced.
Confirmation of this situation in eastern Africa was later provided by lake sediments.54 A more detailed semi-quantitative climate reconstruction further supported the occurrence of widespread aridity at this time (Figure 20).55
Figure 20. Anomaly pattern for a composite of the 1830s, expressed on a seven-point scale ranging from −3 (extremely dry) to +3 (extremely wet).
This approach also demonstrated the occurrence of severe drought in the Sahel in the 1640s, 1680s, c. 1738–1756, the 1770s, the 1790s, and c. 1895–1920, in addition to the early-19th-century drought. Evidence of these early drought episodes helped to disprove the celebrated hypothesis of Charney that the drought commencing in 1968 was of anthropogenic origin due to extensive desertification.56
A Semi-Quantitative Continental Reconstruction since 1800
Several characteristics of historical and geographical documentary material present a challenge to climatic reconstruction. Two of the most serious are the transient and temporary nature of most material. Information derives from widespread locations and is generally not continuous in time.
Figure 21. Map of ninety African regions that form the base of the historical data set.
Moreover, most sources are merely descriptive rather than quantitative. Nicholson established a methodology that dealt with these challenges.57 Ultimately, a semi-quantitative annual precipitation chronology, commencing in 1800, was created for ninety regions of Africa (Figure 21).
The basic methodology involves establishing for each 19th-century entry a time, a location, and a climatological interpretation; determining the larger-scale climatological region for which it can be assumed representative; and converting the entry to a digital value indicative of rainfall conditions at the time represented by the entry. The key to producing a semi-quantitative data set from the raw documentary information was converting this information to individual entries, analogous to an annual rainfall total for a given year and station. This is essentially an analog-to-digital conversion, with the degree of difficulty of the conversion and the accuracy of the entry depending on the nature and amount of detail given in the “raw data.”
A critical aspect of the methodology was establishing geographical regions that are relatively homogeneous with respect to year-to-year changes in precipitation. Thus, in each of the ninety regions shown in Figure 21, it is assumed that an indicator of climate or weather is valid throughout the region. This allows one to assemble into one time series information that comes from a variety of locations. Although this assumption is not perfect, it generally holds for major events such as droughts. This homogeneity of year-to-year rainfall fluctuations over southern Africa is evident in the time series shown in Figure 13. The top two time series show remarkably similar variations, with a correlation of 0.72 over this 105-year period. The correlation is even higher among neighboring regions within South Africa.
The second key to the reconstruction was to assign an index value to each point of information. These ranged from −3 (extremely dry) to +3 (extremely wet), with 0 indicating normal conditions. These values are admittedly subjective, but several criteria were used to assure some consistency among entries. For example, a −2 required a specific mention of drought or of human consequences accompanying a dry year. A −3 generally required a “superlative,” for example, driest year in human memory, or an extremely widespread event.
The first step was to log the relevant information for each entry. An example of entries for Lesotho in several years is shown in Table 2. The information from top to bottom includes region number, index value, geographical region, year referred to, and finally the information obtained from the source. Theses entries are available from the NOAA Paleoclimate Data Center. Most cover the 19th century, but some earlier entries are included. This archive includes a total of 1,915 documentary entries.
The next step was to produce regional averages from these records and then to merge then with whatever gauge data were available. This required the conversion of precipitation measurements to the seven-class index system. The weight given to documentary versus gauge data depended on the number of entries and number of gauges in a given region and year.58 Notably, a record of a large-scale drought in a given year might be more representative than the amount of measured precipitation at a single station, because of the large, random spatial variability of rainfall.
The final step was enhancing spatial detail in the data set. This step was twofold. First, if data were missing for a region and year, but available for a nearby, well-correlated region, the value of the latter was assigned to fill in the blank. After this step a statistical climatic reconstruction method was applied, based on principal component analysis. 59
The resultant 100 x 90 matrix of precipitation index values is shown in Figure 22. Yellow, orange, and red indicate dry conditions; blues and greens indicate wet conditions. Pink entries denote near-normal conditions. Several tendencies emerge from the matrix.
The first is aridity throughout most of the continent in the 1820s and 1830s, and in much of the continent in the 1800s. From the 1840s to c. 1880, a large proportion of regions are indicated as having normal conditions, but some tendency is evident for wetter conditions in equatorial regions (roughly, regions 23 to 50) and drier conditions in the subtropics (e.g., roughly regions 9 to 22, 51 to 68). The 1880s saw relatively good conditions in much of northern Africa but dry conditions in equatorial regions.
Cautionary Notes and Conclusions
The documentary evidence available from historical and geographical sources is a tremendous source of information on past African climates. The information available for Africa has been used to extend the climatic record by at least a century. However, documentary evidence from the past must be interpreted with caution.60 A historical observation is biased by the experiences of the observer. “An extremely dry year” may mean one thing to a European adventurer in Africa and quite another to a local resident. Descriptions of anomalies, such as drought, represent departures from a past—and generally elusive—set of mean conditions. A famine can be provoked by drought, but also by warfare or poor economic conditions, even by flooded fields.
Hydrologic information, be it quantitative or descriptive, can also be complicated to interpret. The Nile flow recorded at the Rodah gauge, for example, is influenced by rainfall in equatorial regions and over the Ethiopian highlands, as well as by the level of Lake Victoria (which determines lake discharge and hence White Nile flow).61 The response of a lake to changes in precipitation depends in part on the lake’s geometry, as well as the relative balance of inflow, outflow, and evaporation and precipitation over the lake itself. Lake Chad, for example, receives inflow from equatorial rivers but lies in the subtropical latitudes of the Sahel. Hence, there is some ambiguity in its interpretation.
Because of the uncertainties and subjective nature of most information related to past climate, the most reliable reconstructions should be based on what might be termed “convergence of evidence.” In other words, are there multiple indicators that suggest the same scenario? This principle is particularly useful when evidence from several disciplines “converges,” that is, lake sediments confirming a drought that is recorded by a few rain gauges.
For Africa, much more documentary information exists that has never been exploited. Most of it, however, is not readily available to the nonspecialist. Retrieval from historical archives and general evaluation of material are generally best done by a historian. On the other hand, climatic interpretation may be best done by the climatologist. Collaboration between the disciplines would be the optimal approach.
Discussion of the Literature
The study of African climate in historical times can be traced back to at least the early 20th century, but climate information has been recorded much longer. Many meteorological records commenced early in the 19th century and were published by national meteorological services or the governments of colonial powers.62 A handful of publications provide inventories or archives of the material related to precipitation.63 In many parts of the continent, oral traditions describe the occurrence of famine and drought over many centuries. This material has often been later summarized in chronologies, such as that of Cissoko.64
With the exception of the meteorological records, climate information was generally sporadic. Continuous records from a set location were rare. One exception is the record of the Nile floods commencing in the year 622 ce.65 Other rivers and lakes also provided information useful for reconstructing African climate history.66 Much of the early material, such as that published by Brooks or Pejml,67 was based on population dynamics, that is, the establishment and decline of empires, the population of oases, caravan routes, or the migration of nomadic societies.
Some of the first systematic attempts at reconstructing African climate history consisted of localized famine and drought chronologies. These were often created for the purpose of understanding such societal issues as disease (e.g., Dias), the slave trade, or economics.68 Sieger’s compilation of lake levels intended to test the climate-sunspot theory.69 Others collected information related to perceived desiccation.70
Perhaps the first works to synthesize material for large sectors of the continent were those of Pejml, Plote, and Nicholson.71 The last source represents a milestone in that it was probably the first attempt to quantify drought and wet years from African historical data, using a seven-class system to classify the degree of wetness of each year. This paper was also the first to provide spatial detail to African historical climate, by identifying continental-scale periods of relatively dry or wet conditions. Nicholson then published the first methodology for African historical climate reconstruction,72 emphasizing the need for “convergence of evidence.” Vogel, Nash, Nash and Endfield, Kelso and Vogel, Grab and Nash, Norrgård, and others followed suit, producing famine and drought chronologies for areas of southern and West Africa.73
The application of statistical methodologies to African historical climate studies has tremendously advanced the spatial and temporal detail, quantification, and completeness of historical records of African rainfall. Nicholson utilized a statistical approach to combine documentary and gauge data and to extend the record to data-poor regions.74 Nicholson and Yin utilized Lake Victoria records to estimate mean rainfall in the lake’s catchment for several periods of the late 18th and 19th centuries.75 Neukom et al. developed a novel approach to combining a group of disparate sources and data types, facilitating the spatial extension and quantification of documentary sources.76 Nicholson et al. used climate reconstruction techniques, based on principal-component analysis, to create a semi-quantitative rainfall data set for Africa.77 It consisted of a “wetness” index (ranging from −3 to +3) for ninety regions of the continent and covering the entire 19th century. Other advances in historical climatology related to Africa are summarized in Nash and Adamson.78
Parallel to the advances in historical climatology are extensions of the precipitation gauge records to earlier decades of the 19th century. This was achieved through two steps. The first was exhaustive searches for rainfall measurements made by official government agencies or individuals. The second was judicious combination of records falling within regions that are relatively homogeneous with respect to inter-annual variability. This resulted in the publication of a gauge record for the Sahel extending from 1854 to 2014 and a similar record for the Guinea Coast extending from 1886 to 2014.79 Similar records, some commencing in the 1830s, were derived for eleven other sectors of the continent, ranging from the Mediterranean coastal regions, through equatorial Africa, and southward to the Cape region of South Africa.80
Several avenues of further research promise to be very fruitful. One is the use of Arabic language sources, would could provide tremendous information for the Sahel, especially for pre-18th century.81 Similarly, there is vast untapped material in missionary records for many areas of the continent. What remains to be done is to compare the current historical chronologies with global indices that can help to ascertain causes of variability and the degree to which they are robust in time. The historical record can also be used to provide analogs for future climates that might prevail under global-warming conditions and to ascertain whether current rainfall variability and extremes may or may not be unprecedented.
Acknowledgment
This work was supported in part by a grant from the National Science Foundation (#AGS1445605). The author would like to thank Douglas Klotter for his work on the figures and manuscript.
Further Reading
- Awange, J. L., V. G. Ferreita, E. Forootan, et al. “Uncertainties in remotely sensed precipitation data over Africa.” International Journal of Climatology 36 (2016): 303–323.
- Brooks, C. E. P. Climate through the Ages. London: Ernest Benn Limited, 1926.
- Brooks, C. E. P., and P. A. Buxton. Le Climat du Sahara et de l’Arabie. Paris: Société d’éditions géographiques, maritimes et coloniales, 1932.
- Grab, S. W., and D. J. Nash. “Documentary Evidence of Climate Variability during Cold Seasons in Lesotho, Southern Africa, 1833–1900.” Climate Dynamics 34 (2010): 473–499.
- Herring, R. “Hydrology and chronology: The Rodah Nilometer as an aid in dating interlacustrine history.” In Chronology, Migration and Drought in Interlacustrine Africa. Edited by J. B. Webster, 38–86. Halifax: Dalhousie University Press, 1979.
- Johnston, H. H. British Central Africa: an Attempt to Give Some Account of a Portion of the Territories under British Influence North of the Zambesi, 544. New York: Edward Arnold, 1897.
- Lamb, H. H. Climate: Present, Past and Future. Vol. 2, Climatic History and the Future. London: Methuen, 1977.
- Nash, D. J., and G. C. D. Adamson. “Recent Advances in the Historical Climatology of the Tropics and Subtropics.” Bulletin of the American Meteorological Society 42 (2014): 131–146.
- Nash, D. J., and G. H. Endfield. “Historical Flows in the Dry Valleys of the Kalahari Identified from Missionary Correspondence.” South African Journal of Science 98 (2002): 244–248.
- Nash, D. J., et al. “African Hydroclimatic Variability during the Last 2000 Years.” Quaternary Science Reviews 154 (2016): 1–22.
- Nicholson, S. E. “The Methodology of Historical Climate Reconstruction and Its Application to Africa.” Journal of African History 20 (1979): 31–49.
- Nicholson, S. E. “Environmental Change within the Historical Period.” In The Physical Geography of Africa. Edited by A. S. Goudie, W. M. Adams, and A. Orme, 60–75. Oxford: Oxford University Press, 1996.
- Nicholson, S. E. “A Semi-Quantitative, Regional Precipitation Data Set for Studying African Climates of the Nineteenth Century, Part 1: Overview of the Data Set.” Climatic Change 50 (2001): 317–353.
- Nicholson, S. E., A. K. Dezfuli, and D. Klotter. “A Two-Century Precipitation Dataset for the Continent of Africa.” Bulletin of the American Meteorological Society 93 (2012): 1219–1231.
- Nicholson, S. E., D. Klotter, and A. K. Dezfuli. “Spatial Reconstruction of Semi-Quantitative Precipitation Fields over Africa during the Nineteenth Century from Documentary Evidence and Gauge Data.” Quaternary Research 78 (2002): 13–23.
- Nicholson, S. E., D. J. Nash, B. M. Chase, S. W. Grab, T. M. Shanahan, A. Asrat, A.-M. Lézine, D. Verschuren, and M. Umer. “Temperature Variability over Africa during the Last 2,000 Years.” The Holocene 23 (2013): 1085–1094.
- Pejml, K. Studie o Kolísání Klimatu v Severní Africa v Prehistorickych a Historickych Dobach. Prague: Praha, 1963.
- Plote, G. L’Afrique Sahélienne se dessèche-telle? Orleans, France: BRGM, Département géologie de I'Aménagement, 1974.
- Trouet, V., K. Haneca, P. Coppin, and H. Beeckman. “Tree ring analysis of Brachystegia spiciformis and Isoberlinia tomentosa: Evaluation of the ENSO- signal in the miombo woodland of eastern Africa.” International Association of Wood Anatomists Journal 22 (2001): 385–399.
- Weatherby, J. “The raindrums of the Sor.” In Chronology, Migration and Drought in Interlacustrine Africa. Edited by J. B. Webster, 317–323. New York: Africana, 1979.
Notes
2. C. Ballard, “Drought and Economic Distress: South Africa in the 1800s,” Journal of Interdisciplinary History 17 (1986): 359–378.
3. A. Dalzel, The History of Dahomey (London: Frank Cass, 1793); and I. A. Akinjogbin, Dahomey and Its Neighbors, 1708–1818 (Cambridge, U.K.: Cambridge University Press, 1967).
4. D. Winstanley, “Recent Rainfall Trends in Africa, the Middle East and India,” Nature 243 (1973): 464–465; and A. H. Bunting, M. D. Dennette, J. Elston, and J. R. Milford, “Rainfall Trends in the West African Sahel,” Quarterly Journal of the Royal Meteorological Society 102 (1976): 59–64.
5. U. Schneider, A. Becker, P. Finger, A. Meyer-Christoffer, B. Rudolf, and M. Ziese, “GPCC Full Data Reanalysis Version 7.0 at 1.0°: Monthly Land-Surface Precipitation from Rain-Gauges Built on GTS-based and Historic Data, 2015.”
6. M. Hulme, “A 1951–80 Global Land Precipitation Climatology for the Evaluation of General Circulation Models,” Climate Dynamics 7 (1992): 57–72; and M. Hulme, T. J. Osborn, and T. C. Johns, “Precipitation Sensitivity to Global Warming: Comparison of Observations with HadCM2 Simulations,” Geophysical Research Letters 25 (1998): 3379–3382.
7. C. J. Willmott, and K. Matsuura, “Terrestrial Air Temperature and Precipitation: Monthly and Annual Time Series (1950–1999),” Version 1.02, 2001.
9. G. P. Compo et al., “The Twentieth Century Reanalysis Project,” Quarterly Journal of the Royal Meteorological Society 137 (2011): 1–28.
10. M. Cremaschi, M. Pelfini, and M. Santilloi, “Cupressus Dupreziana: A Dendroclimatic Record for the Middle-Late Holocene in the Central Sahara,” The Holocene 16 (2006): 293–303; and R. Touchan, D. M. Meko, and A. Aloui, “Precipitation Reconstruction for Northwestern Tunisia from Tree Rings,” Journal of Arid Environments 72 (2008): 1887–1896.
11. D. J. Nash et al., “African Hydroclimatic Variability during the Last 2000 Years,” Quaternary Science Reviews 154 (2016): 1–22.
12. A. S. Cohen, M. R. Talbot, S. M. Awramik, D. L. Dettman, and P. Abell, “Lake Level and Paleoenvironmental History of Lake Tanganyika, Africa as Inferred from Late Holocene and Modern Stromatolites,” Geological Society of America Bulletin 109 (1997): 444–460; and J. C. Stager, C. Cocquyt, R. Bonnefille, C. Weyhenmeyer, and N. Bowerman, “A Late Holocene Paleoclimatic History of Lake Tanganyika, East Africa,” Quaternary Research 72 (2009): 47–56.
13. T. M. Shanahan, J. T. Overpeck, J. W. Beck, C. W. Wheeler, J. A. Peck, J. W. King, and C. A. Scholz, “The Formation of Biogeochemical Laminations in Lake Bosumtwi, Ghana, and Their Usefulness as Indicators of Past Environmental Changes,” Journal of Paleolimnology 40 (2008): 339–355; C. Wolff, G. H. Haug, A. Timmermann, J. S. S. Damsate, A. Brauer, D. M. Sigman, M. A. Cane, and D. Verschuren, “Reduced Interannual Rainfall Variability in East Africa during the Last Ice Age,” Science 333 (2011): 743–747; S. L. Barry, M. L. Filippi, M. R. Talbot, T. C. Johnson, “A 20000-Yr Sedimentological Record from Lake Malawi, East Africa: The Late-Pleistocene/Holocene Transition in the Southern Tropics,” in The East African Great Lakes: Limnology, Paleoclimatology and Biodiversity, ed. E. O. Odada and D. O. Olago (Dordrecht, The Netherlands: Kluwer Academic, 2002), 369–392.
14. S. Mulitza et al., “Increase in Africa Dust Flux at the Onset of Commercial Agriculture in the Sahel Region,” Nature 466 (2010): 226–228.
15. Nash et al., “African Hydroclimatic Variability.”
16. M. D. Therrell, D. W. Stahle, L. P. Ries, and H. H. Shugart, “Tree-Ring Reconstructed Rain-Fall Variability in Zimbabwe,” Climate Dynamics 26 (2006): 677–685; Trouet et al. (2001); and E. Fichtler, V. Trouet, H. Beeckman, P. Coppin, and M. Worbes, “Climatic Signals in Tree Rings of Burkea africana and Pterocarpus angolensis from Semiarid Forests in Namibia,” Trees Structure and Function 18 (2004): 442–451.
17. P. W. Dunwiddie, and V. C. LaMarche, “A Climatically Responsive Tree-Ring Record from Widdringtonia cedarbergensis, Cape Province, South Africa,” Nature 286 (1980): 796–797.
18. S. E. Nicholson, D. J. Nash, B. M. Chase, S. W. Grab, T. M. Shanahan, A. Asrat, et al., “Temperature Variability over Africa during the Last 2,000 Years,” The Holocene 23 (2013): 1085–1094.
19. J. E. Tierney, M. T. Mayes, N. Meyer, C. Johnson, P. W. Swarzenski, A. S. Cohen, and J. M. Russell, “Late-Twentieth-Century Warming in Lake Tanganyika Unprecedented since AD 500,” Nature Geoscience 3 (2010): 422–425.
20. C. E. P. Brooks, Climate through the Ages (London: Ernest Benn, 1926); Brooks and P. A. Buxton, Le Climat de Sahara et de lArabie (Paris: Société d’éditions géographiques, maritimes et coloniales, 1932); H. H. Lamb, Climate and History, Climate: Present, Past and Future; Volume 2: Climatic History and the Future (London: Methuen, 1977); and Emmanuel Le Roy Ladurie, Histoire du climat depuis l’an mil (Paris: Flammarion, 1967).
21. K. Pejml, Studie o Kolísání Klimatu v Severní Africa v Prehistorickych a Historickych Dobach. Hydrometeorologicky Ustav (Prague: Studies of Climatic Fluctuations in Northern Africa in Prehistoric and Historic Epochs, 1963).
22. G. Plote, L’Afrique Sahélienne se dessèche-telle? (Orleans, France, 1974).
23. A. Danckelmann, Mémoire sur les observations météorologiques faites à Vivi et sur la climatologie de la côte Sud-Ouest d’Afrique en général (Berlin: Asher, 1884).
24. H. Barth, Travels and Discoveries in North and Central Africa (1849–1855): Being a Journal of an Expedition Undertaken under the Auspices of H. B. M.’s Government, in the Years 1849–1855 (London: Frank Cass, 1965).
25. A. Borius, Recherches sur le climat du Sénégal (Paris: Gauthier Villars, 1875); A. Borius, Nouvelles recherches sur le climat du Sénégal (Paris: Gauthier Villars, 1879); A. Borius, Recherches sur le climat des établissements français de la côte septentrionale du Golfe de Guinée (Paris: Gauthier Villars, 1880); V. Raulin, Observations pluviometriques faites dans l’Algérie, 1751–1870 (Paris: Savy, 1876); and R. Rousseau, “Les pluies au Sénégal de 1887 à 1927,” Bulletin du Comité d’études historiques et scientifiques de l’A.O.F. 14.1–2 (1935): 157–182.
26. S. E. Nicholson, “The Methodology of Historical Climate Reconstruction and Its Application to Africa,” Journal of African History 20 (1979): 31–49; S. E. Nicholson, A. K. Dezfuli, and D. Klotter, “A Two-Century Precipitation Dataset for the Continent of Africa,” Bulletin of the American Meteorological Society 93 (2012): 1219–1231; and S. E. Nicholson, D. Klotter, and A. K. Dezfuli, “Spatial Reconstruction of Semi-Quantitative Precipitation Fields over Africa during the Nineteenth Century from Documentary Evidence and Gauge Data,” Quaternary Research 78 (2012): 13–23.
27. S. E. Nicholson, “A Semi-Quantitative, Regional Precipitation Data Set for Studying African Climates of the Nineteenth Century, Part 1: Overview of the Data Set,” Climatic Change 50 (2001): 317–353.
28. V. Monteil, “Chroniques du Tichitt,” Bulletin de l’ Institut Fondamental d’ Afrique Noire 1 (1939): 282–313; P. Marty, Poème historique d’Abou Bakr Ibn Hejab, le Dimani, Bulletin du Comité d’études historiques et scientifiques de l’Afrique occidentale française 4 (1921): 252–263; V. Monteil, Chronique du Wâlo Empire (1186–1855), Bulletin de l’ Institut Fondamental d’ Afrique Noire 26 (1964): 440–498; S.-M. Cissoko, Famines et épidemies à Tombouctoo et dans la Boucle du Niger du XVIe au XVIIIe siècle, Bulletin de l’ Institut Français de l’Afrique Noire 3B (1968): 806–821; and Y. Urvoy, Histoire de l’empire du Bornu, Memoires de l’ Institut Français de l’Afrique Noire 7 (1949): 1–166.
29. J. B. Webster, “Noi! Noi! Famines as an Aid to Interlacustrine Chronology,” in Chronology, Migration and Drought in Interlacustrine Africa, ed. J. B. Webster (Halifax, NS: Dalhousie University Press, 1979), 1–37. Herring (1979).
30. R. B. Owen, R. Crossley, T. C. Johnson, D. Tweddle, I. Kornfield, S. Davison, et al., “Major Low Levels of Lake Malawi and Their Implications for Speciation Rates in Cichlid Fishes,” Proceedings of the Royal Society of London 240 (1990): 519–553.
31. P. Shaw, “A Historical Note on the Outflows of the Okavango Delta System,” Botswana Notes and Records 16 (1984): 127–130; S. E. Nicholson, “Environmental Change within the Historical Period,” in The Physical Geography of Africa, ed. A. S. Goudie, W. M. Adams, and A. Orme (Oxford: Oxford University Press, 1996), 60–75; S. E. Nicholson, “Fluctuations of Rift Valley Lakes Malawi and Chilwa during Historical Times: A Synthesis of Geological, Archaeological and Historical Information,” in Environmental Change and Response in East African Lakes, ed. J. T. Lehman (Dordrecht, The Netherlands: Kluwer, 1998), 207–231; and S. E. Nicholson, “Historical and Modern Fluctuations of Lakes Tanganyika and Rukwa and Their Relationship to Rainfall Variability,” Climatic Change 41 (1999): 53–71.
33. Owen et al., “Major Low Levels.”
34. J. Maley, “Mécanisme des changements climatiques aux basses latitudes,” Palaeogeogr. Palaeoclimatol. Palaeoecol 14 (1973): 193–227; and Nicholson, “Environmental Change.”
35. O. Toussoun, “Memoire sur l’histoire du Nil,” Memoires de l’Institut d’Egypt 18 (1925): 366–404.
36. R. Sieger, “Schwankungen der innerafrikanischen Seen,” Verein der Geographie, Jahres bericht (Vienna: University of Vienna, 1887), 41–60.
37. S. E. Nicholson, and D. Entekhabi, “The Quasi-Periodic Behavior of Rainfall Variability in Africa and Its Relationship to the Southern Oscillation,” Archives for Meteorology, Geophysics and Bioclimatology, Series A, 34 (1986): 311–348; and H. Rodhe and H. Virji, Trends and Periodicities in East African Rainfall. Monthly Weather Review 104 (1976): 307–315.
38. I. A. Lund, “Map-Pattern Classification by Statistical Methods,” Journal of Applied Meteorology 2 (1963): 56–65; T. J. Blasing, “Methods for Analyzing Climatic Variations in the North Pacific Sector and Western North America for the Last Few Centuries” (PhD diss. University of Wisconsin, Madison, 1975); and S. E. Nicholson, “The Nature of Rainfall Fluctuations in Subtropical West Africa,” Monthly Weather Review 108 (1980): 473–487.
39. L. S. Kalkstein, G. Tan, and J. A. Skindlov, “An Evaluation of Three Clustering Procedures for Use in Synoptic Climatological Classification,” Journal of Climate and Applied Meteorology 26 (1987): 717–730.
40. A. K. Dezfuli, “Spatio-Temporal Variability of Seasonal Rainfall in Western Equatorial Africa,” Theoretical and Applied Climatology 104 (2011): 57–69.
41. M. E. Mann and P. D. Jones, “Global Surface Temperatures over the Past Two Millennia,” Geophysical Research Letters 30 (2003): 1820; R. Neukom et al. “Multiproxy Summer and Winter Surface Air Temperature Field Reconstructions for Southern South America Covering the Past Centuries,” Climate Dynamics (2014); and J. E. Smerdon, A. Kaplan, D. Chang, and M. N. Evans, “A Pseudoproxy Evaluation of the CCA and RegEM Methods for Reconstructing Climate Fields of the Last Millennium,” Journal of Climate 24 (2011): 1285–1309.
42. S. E. Nicholson, “A Climatic Chronology for Africa: Synthesis of Geological, Historical and Meteorological Information and Data” (PhD diss. University of Wisconsin, Madison, 1976); S. E. Nicholson, “Climatic Variations in the Sahel and Other African Regions during the Past Five Centuries,” Journal of Arid Environments 1 (1978): 3–24; S. E. Nicholson, “Saharan Climates in Historic Times,” in The Sahara and the Nile, ed. M. A. J. Williams and H. Faure (Rotterdam, The Netherlands: A. A. Balkema, 1980), 173–200; George Brooks, “A Provisional Historical Schema for Western Africa Based on Seven Climatic Period,” Cahiers d’études africaines 26 (1986): 43–62.
43. S. E. Nicholson, “The Historical Climatology of Africa,” in Climate and History, ed. T. M. L. Wigley, M. J. Ingram, and G. Farmer, 249–270 (Cambridge, U.K.: Cambridge University Press, 1981).
44. Nicholson, “A Semi-Quantitative, Regional Precipitation Data Set”; Nicholson et al., “A Two-Century Precipitation Dataset”; and Nicholson et al., “Spatial Reconstruction.”
45. C. Vogel, “A Documentary-Derived Climatic Chronology for South Africa, 1820–1900,” Climatic Change 14 (1989): 291–308; C. Kelso, and C. Vogel, “The Climate of Namaqualand in the Nineteenth Century,” Climatic Change 83 (2007): 357–380; D. J. Nash, “The Dry Valleys of the Kalahari: Documentary Evidence of Environmental Change in Central Southern Africa,” Geographical Journal 162 (1996): 154–168; D. J. Nash and G. H. Endfield, “Historical Flows in the Dry Valleys of the Kalahari Identified from Missionary Correspondence,” South African Journal of Science 98 (2002): 244–248; D. J. Nash and G. H. Endfield, “‘Splendid Rains Have Fallen’: Links between El Niño and Rainfall Variability in the Kalahari, 1840–1900,” Climatic Change 86 (2008): 257–290; D. J. Nash and S. W. Grab, “‘A Sky of Brass and Burning Winds’: Documentary Evidence of Rainfall Variability in the Kingdom of Lesotho, Southern Africa, 1824–1900,” Climatic Change 101 (2010): 617–653; D. J. Nash et al. “Tropical Cyclone Activity over Madagascar during the Late Nineteenth Century,” International Journal of Climatology 35 (2015): 3249–3261; S. W. Grab and D. J. Nash, “Documentary Evidence of Climate Variability during Cold Seasons in Lesotho, Southern Africa, 1833–1900,” Climate Dynamics 34 (2010): 473–499; and Neukom et al., “Multiproxy Summer and Winter Surface Air Temperature.”
46. M. J. Hannaford, J. M. Jones, and R. Grant, “Early-Nineteenth-Century Southern African Precipitation Reconstructions from Ships’ Logbooks,” Holocene 25 (2015): 379–390.
47. D. Anderson, “The Beginning of Time? Evidence for Catastrophic Drought in Baringo in the Early Nineteenth Century,” Journal of Eastern African Studies 10 (2016): 45–66.
48. S. Norrgård, A New Climatic Periodisation of the Gold and Guinea Coasts in West Africa, 1750–1798 (Åbo, Finland: Åbo Akademi University Press, 2013); and S. Norrgård, “Practising Historical Climatology in West Africa: A Climatic Periodisation 1750–1800,” Climatic Change 129 (2015): 131–143.
49. Nicholson, “Climatic Variations”; Nicholson, “Saharan Climates”; Nicholson, “The Historical Climatology”; and Nicholson, “Environmental Change.”
51. R. R. K. Pankhurst, An Introduction to the Economic History of Ethiopia (London: Laibela House, 1961); R. R. K. Pankhurst, Economic History of Ethiopia (Addis Ababa: Haile Selassie 1 University Press, 1968); and J. Miller, “The Significance of Drought, Disease and Famine in the Agriculturally Marginal Zones of West-Central Africa,” Journal of African History 23 (1982): 17–61.
52. Nicholson, “Climatic Variations.”
53. M. L. Abdelhadi, “Analyse de La Sécheresse Qui a Sevi Au Maroc de 1980 à 1985, Cas Du Bou Regreg” (Rabat, Morocco: Departmente d’Aménagements Hydrauliques, 1987); Raymond A. Almeida, “Chronological References: Cabo Vere/Cape Verdean American,” 1997. C. Becker, “Note Sur Les Conditions écologiques En de La Sénégambie Au 17e et 18e Siècle,” African Economic History 14 (1985): 167–216; George E. Brooks “Cabo Verde: Gulag of the South Atlantic: Racism, Fishing Prohibitions, and Famines,” History in Africa 33 (2006): 101–135; Jean Marchika, La Peste en Afrique Septentrionale: Histoire de la Peste en Algérie de 1363 à 1830 (Algiers, Algeria: University of Algiers, 1927); S. E. Nicholson, “Historical Fluctuations of Lake Victoria and Other Lakes in the Northern Rift Valley of East Africa,” in Environmental Change and Response in East African Lakes, ed. J. T. Lehman (Dordrecht, The Netherlands: Kluwer, 1998), 7–35; Nicholson, “Historical and Modern Fluctuations of Lakes Tanganyika”; and K. D. Patterson, “Epidemics, Famines, and Population in the Cape Verde Islands, 1580–1900,” International Journal of African Historical Studies 21 (1988): 291–313.
54. D. Verschuren, K. R. Laird, and B. F. Cumming, “Rainfall and Drought in Equatorial East Africa during the Past 1,100 Years,” Nature 403 (2000): 410–414.
55. Nicholson et al., “A Two-Century Precipitation Dataset”; Nicholson et al., “Spatial Reconstruction.”
56. J. G. Charney, “Dynamics of Deserts and Drought in Sahel” Quarterly Journal of the Royal Meteorological Society 101 (1975): 193–202.
57. Nicholson, “A Semi-Quantitative, Regional Precipitation Data Set.”
58. Nicholson et al., “Spatial Reconstruction.”
59. Nicholson et al., “Spatial Reconstruction.”
60. Nicholson, “The Methodology of Historical Climate Reconstruction.”
61. Nicholson, “Historical Fluctuations”; and S. E. Nicholson, X. Yin, and M. B. Ba, “On the Feasibility of Using a Lake Water Balance Model to Infer Rainfall,” Hydrological Sciences Journal 45 (2000): 75–95.
62. T. C. Peterson and J. F. Griffiths, “Historical African Data,” Bulletin of the American Meteorological Society 78 (1997): 2869–2872.
63. For example, A. Supan, “Die Verteiliung des Niederschlags auf der festen Erdoerfläche,” Petermanns Mitteilungen, Ergänzungsheft 124, 1898; and V. Raulin, Observations pluviometriques faites dans l’Algérie, 1871–1880 (Paris: Savy, 1882).
64. S.-M. Cissoko, “Famines et épidemies à Tombouctoo et dans la Boucle du Niger du XVIe au XVIIIe siècle,” Bulletin de l’ Institut Français de l’Afrique Noire 3B (1968): 806–821.
65. Toussoun, “Memoire sur l’histoire du Nil.”
66. Sieger, “Schwankungen der innerafrikanischen Seen”; Maley, “Mécanisme des changements climatiques”; and Shaw, “A Historical Note.”
67. Brooks, Climate through the Ages; and K. Pejml, “A Contribution to the Historical Climatology of Morocco and Mauritania,” Studie Geophysica Geodaetica 6 (1962): 257–280.
68. For example, Miller, “The Significance of Drought”; and Pankhurst, Economic History.
69. Sieger, “Schwankungen der innerafrikanischen.”
70. E. H. L. Schwarz, The Kalahari or Thirstland Redemption (Cape Town: Maskew Miller, 1920).
71. Pejml, Studie o Kolísání Klimatu; Plote, L’Afrique Sahélienne; and Nicholson, “A Climatic Chronology.”
72. Nicholson, “The Methodology of Historical Climate.”
73. Vogel, “A Documentary-Derived Climatic Chronology”; Nash, “The Dry Valleys of the Kalahari”; Nash and Endfield, “Historical Flows in the Dry Valleys”; Nash and Endfield, “‘Splendid Rains Have Fallen’”; Kelso and Vogel, “The Climate of Namaqualand”; Grab and Nash, “Documentary Evidence of Climate Variability”; and Norrgård, “Practising Historical Climatology.”
74. Nicholson, “A Semi-Quantitative, Regional Precipitation Data Set.”
75. S. E. Nicholson and X. Yin, “Rainfall Conditions in Equatorial East Africa during the Nineteenth Century as Inferred from the Record of Lake Victoria,” Climatic Change 48 (2001): 3878–3898.
76. Neukom et al. “Multiproxy Summer and Winter Surface.”
77. Nicholson et al., “A Two-Century Precipitation Dataset”; and Nicholson et al., “Spatial Reconstruction of Semi-Quantitative Precipitation.”
78. D. J. Nash and G. C. D. Adamson, “Recent Advances in the Historical Climatology of the Tropics and Subtropics,” Bulletin of the American Meteorological Society 95 (2014): 131–146.
79. S. E. Nicholson, A. Fink, and C. Funk, “Overview of Rainfall Variability in the Sahel and Guinea Coast since the Mid-Nineteenth Century,” International Journal of Climatology, 2017, submitted.
80. S. E. Nicholson, C. Funk, and A. Fink, “One and a Half Centuries of Rainfall Variability over the African Continent,” Global and Planetary Change, 2017, submitted.
81. Nash and Adamson, “Recent Advances in the Historical Climatology.”
