The term dryland is a collective noun for all types of arid region. Drylands cover over one third of the present global land area, which rises to nearly 50% if the dry sub-humid category is included. Such regions are highly variable in climate. The term 'desert' is a botanical description, and is avoided because it is inadequately defined.

There are many schemes to classify aridity of climate zones, each with different end-uses such as food production, water-supply, geomorphology. In essence they all consider moisture availability, and look at the balance between supply (precipitation) and losses (evaporation and transpiration from plants).

Arid zones are those where evapotranspiration exceeds precipitation. It is important to keep in mind that amount of rainfall alone is not the absolute control. For example, in the Lake Eyre basin of central Australia, annual precipitation may be up to 300mm, but losses commonly exceed 3000mm of rainfall equivalent (Kotwicki & Allan, 1998).

Some schemes try to take into account retention of moisture in soils, since this affects how rapidly rainfall is converted into runoff. Some try and account for seasonality of precipitation, since the effect of rainfall spread evenly through the year is clearly going to be different from the same rainfall arriving in just two months of the year. Yet others take into account the mean summer or winter temperatures, or the difference between them, since this controls the variation in evapo-transpiration throughout the year.

In practice, nearly all the schemes end up classifying the same parts of the planet in essentially the same way. All have subdivisions into semi-arid, arid, and hyper-arid, based on a scale of increasing aridity. Some schemes include a slightly less arid zone called dry sub-humid.

The widely applied United Nations classification (UNEP, 1992) uses an aridity index, AI, where AI = P/PET. P is annual precipitation. PET is potential evapotranspiration, which is calculated from meteorological data. In this scheme, the subdivisions are:

As a rough guide, the subdivisions of aridity can be equated approximately to mean annual precipitation. When this is done, it is crucial to bear in mind that rainfall alone is not the sole control on aridity, and to remember that averaging rainfall is misleading, because in arid zones there may be many years when rainfall is zero. A commonly used conversion is that of Grove (1977):

When studying the rock record, clearly we have no meteorological data available, and have to use proxies instead. We cannot directly estimate either precipitation or evapotranspiration, all we can see is the effect of the balance between them.

Drylands are created where climatic, topographic or oceanographic factors prevent moisture-bearing weather systems reaching that zone.

Global atmospheric patterns
The sub-tropical regions are dominated by dry high-pressure air. Solar heating of the atmosphere is greatest at the Equator, making air rise, which in turn causes it to cool, producing water condensation and creating a cloudy central zone with equatorial belts of rain. The air spreads out at high altitude towards the poles and descends in the subtropics at around 30° north and south. The now-dry air warms adiabatically as it descends, creating a consistently warm and dry region. The air returns near the surface to the tropics, as the Trade Winds, picking up moisture on the way. This stable tropical-to-subtropical circulation is called a Hadley Cell.

At higher latitudes, air circulation is controlled more by the rotation of the Earth, and by the temperature contrast between the Equator (most solar radiation) and the poles (least solar radiation). At high altitudes, air generally moves west to east (jet streams), and warm moist air is carried towards the poles. Cold and dry polar air moves back again at lower altitudes, interacting as it goes with the polewards moving warm air. The outcome is a complex pattern of mobile weather systems (cyclones and anticyclones) displaying considerable vertical and lateral instability. In general, this high-latitude pattern is not conducive to creating aridity.

This global circulation pattern means that, in general, the subtropics are warm arid zones where precipitation is scarce and irregular. These drylands are surrounded by narrow semi-arid regions. The main relief in such regions comes from seasonal monsoons penetrating the high-pressure cells from the Equator.
The diffuse boundary between the equatorial Hadley cells and the more energetic high-latitude weather systems is called the Inter-Tropical Convergence Zone (ITCZ). This is the zone of maximum aridity. The position of the ITCZ moves up to about 20° south during the northern-hemisphere winter, and north again during the summer, because the tilt of the Earth's axis means the zone receiving maximum solar radiation itself moves progressively south then north, respectively, of the Equator during the year. Therefore the zone of maximum aridity moves south and north with the seasons.

Continentality
Distance from marine or other sources of moisture encourages aridity. So drylands occur near the centres of large landmasses. The Trade Winds blow from east to west, so drylands commonly extend westwards to the coasts because the winds lose their moisture in travelling across large continents. This effect produces small arid zones surrounded by broad semi-arid regions.

Topography
Mountain ranges create rain shadows and prevent moisture-laden air from entering some regions by forcing the air to rise, which makes it cool and causes the water to condense and precipitate. On the lee side, the descending air is warmed and dried adiabatically. The Great Divide of eastern Australia cuts off the interior from the Trade Winds. The Andes shelter the Argentinian drylands.

Cold ocean currents
Cold ocean currents flowing from the Poles to the Equator affect the western coastal margins of South America, southern Africa and Australia. These currents reinforce cause low rates of sea-surface evaporation, low precipitation (often as fog and dew due to the lower temperatures), and a low temperature range, reinforcing other climatic factors. For example, the lack of rainfall in the Namib Desert is due to the cold Benguela current impacting the west margin and the failure of easterly rain-bearing winds to penetrate across the African continent.

Fig. 1 shows the present-day distribution of drylands around the planet. North Africa and Arabia are clearly dominated by the global subtropical high pressure belt. Central Asia is dominated by the continentality effect. North American drylands are the combined product of topography (Sierra Nevada and Rocky Mountains) and continentality. South America is dominated by the combined effect of cold ocean current on the west coast, and the elevated topography of the Andes.

An overview of the geomorphology of each of the main drylands on the planet today can be found in Oberlander (1994).

Drylands currently occupy more than one-third of the global land surface. Using the UNEP (1992) definitions, the proportions are:

Annual temperature regimes vary considerably, though a wide temperature range is common to all. Temperature affects the seasonal availability of moisture, and the form in which precipitation arrives (snow in high latitudes). It also affects vegetation types. A common subdivision is (Meigs, 1952):

Boundaries between regions are diffuse, and change with time. On this scheme, examples of hot drylands would be Arabia, central Sahara, central Australia; mild drylands are the Kalahari, southern Sahara, Mexico, northern Australia; cool drylands are northern Sahara, Atacama, and southern California; cold drylands are Gobi, China, Canadian Prairies, and higher parts of the SW USA such as the Colorado Plateau.

The climate of modern drylands is known to have changed over short time-spans. At the scale of decades (10¹ years), fluctuations occur because of short-term instability in global circulation patterns, and ocean-controlled phenomena such as El Niño-Southern Oscillation (ENSO) in the Pacific, and its Atlantic equivalent. These produce extended periods (5-10 years) of drought in some locations while causing extended wet periods elsewhere, and will produce a pulsed sedimentological signature.

At the century time-scale (10² years), the southern margin of the Sahara has migrated southwards in the Sahel region (LeHouerou, 1968), though it is unclear if these changes were triggered by human influence such as over-grazing.

At the thousand-year scale (10³ years), major changes in climate are well documented which are clearly not triggered by human influence. For example, there seems to have been a global increase in aridity at about 4.5 to 3.5 ka BP which coincides with the socio-economic collapse of several cultures (Petit-Maire & Guo, 1998).

We know the Pleistocene glaciations had a marked effect on many drylands, producing changes at the 10⁴-10⁵ year scale which were seen as wetting and drying phases. During the glacials, aridity increased, as moisture was locked up in the ice, and winds were stronger, which increased evaporation rates. In the inter-glacials, there were periods much wetter than the present. Such variation is well documented even in areas far away from glaciated regions, such as in central Australia (Nanson et al., 1992).

But the effects of the glacial periods on depositional environments were not perhaps as one might expect. During the glacials, temperatures were depressed, which reduced evaporation rates. So, even though rainfall was still restricted, lakes became permanent or larger, and rivers flowed more persistently; an effect which has been clearly documented in Lake Eyre (Nanson et al., 1998). This is a phenomenon that would have been relevant even at times when global glaciation was negligible, such as in the Triassic, and again demonstrates that precipitation alone is not the dominant control on drylands.

When considering possible climate changes in the Triassic, it should be noted that computer-based climate modelling suggests that Pangaea was unusually sensitive to orbital (Milankovitch) forcing effects on climate (Crowley, 1994; Veevers, 1994), though it may be southern Pangaea was more sensitive than northern parts. We might therefore expect to see variations with typical Milankovitch periodicities, currently averaging at about 21ka, 41ka, and 100ka (Williams et al., 1993 p.81-85).

At the much longer time scale of 10⁶-10⁷ years (geological periods), major climate shifts took place as continental drift moved drylands into more humid climatic settings, and changes occurred in global atmospheric and oceanic circulation patterns. What is not clear is how the transition from dryland to humid-land would have affected sedimentation. Is there a gradual change, or does the system cross a threshold and rapidly switch into a new behaviour? Geomorphological research over the last three decades would suggest the latter is more likely (e.g. Schumm & Parker, 1974; Begin & Schumm, 1984).

Great care must be exercised when interpreting climate change events from the sedimentological record to use as local or regional correlation markers.

Dryland rivers sediments are not reliable at recording small to medium amounts of climate change. Periods of extreme aridity leave little record: rivers are inactive, the only clue may come from aeolian reworking but this is possible even in wet periods. The preserved record is dominated by deposits from periods of wetter and cooler climate. Even when dryland sedimentary systems do respond to change, the response may not yield an unequivocal record: for example, lakes may form and grow larger, and rivers may have more persistent flow, either because the climate is wetter or because the average temperature has dropped.

Because sedimentary processes in drylands are irregular, as a general rule dryland rivers are out of equilibrium with the local conditions. This contrasts with perennial rivers, in humid climates, which respond quickly to local change. If the sediments of dryland rivers are not correctly recording the ambient conditions, it is highly unlikely they will correctly record changes in the conditions unless the magnitude of change is large and maintained for a long time.

Changes to dryland climate do not have to be large to elicit a dramatic and complex response, if the change forces the river system to cross a geomorphic threshold (Bull, 1979). Indeed, the whole concept of geomorphic thresholds was developed from study of dryland rivers (e.g. Schumm & Hadley, 1957; Patton & Schumm, 1981).

The difficulty for those interpreting the rock record is that it is easily possible to have changes occur which are large enough to produce a river response, yet which can be localised in geographic extent (e.g. restricted to one valley system). For example, a difference in elevation (so changing average temperature) between two valleys may be sufficient to inhibit in one the effects of rainfall reduction seen in the other.

Climate changes occur with a wide variety of periodicities, the rock record is simply the additive response to these different superimposed frequencies. Which will be dominant and so leave the sedimentological record?

Average sedimentation rates for dryland rivers are relatively low, and decrease with increasing aridity. Furthermore, large magnitude but infrequent flood events dominate the sedimentary processes, so the rock record is going to be biased to these rare events. It is highly likely, therefore, that short-period changes in climate will be inaccurately represented, and may leave no record at all.

The signatures of change we are most likely to discern are going to be the large magnitude and long-period events. These are not good for correlation because the transition between phases is long, leaving a low-resolution record (where do you pick the change to have occurred?).

It is the cumulative effect of these problems that leads to disagreement between workers on the recognition of "wetting and drying" cycles in red-bed successions, a phenomenon well know to those working on the Rotliegend Group of the southern North Sea. For the same succession, one worker sees drying-up cycles, another sees them as wetting-up.

It is abundantly clear, despite the above warnings, that drylands have been dominant features of the planet surface in the geological past. We can tell this for particular time periods, and some geographical locations, because of the distinctive sedimentary record left behind. For example, the region currently occupied by the Colorado Plateau, in SW USA, was a persistent dryland from the late Permian to the Middle Jurassic, as reflected by the substantial thickness of aeolian sandstones (e.g. the Cedar Mesa, Wingate, Navajo, and Entrada sandstones - see review by Blakey et al., 1988). There can be no doubt that the fluvial sediments intercalated between these aeolian units were the deposits of dryland rivers.

Modern plate reconstructions allow us to understand the causes of aridity in the past. For example, the Colorado Plateau was in a subtropical position for much of the Mesozoic era. The time of the Pangaean supercontinent provided ample opportunity for the continentality effect to produce huge dryland areas (Veevers, 1994). In other cases, we can deduce the presence of major topographic barriers.

As an example, it is possible to make some general predictions about aridity in the Triassic simply by looking at plate reconstructions. The Lower Triassic of NW Europe is characterised by a variety of aeolian, fluvial and other continental lithofacies characterstic of drylands.

At this time, the Pangaea super-continent had been well established for some time, and was about to break up (Klein et al., 1994). Much of Pangaea experienced an arid climate because of its situation around the sub-tropics (global circulation patterns), and the large size of the land mass (continentality).

In the Early Triassic, the Variscan-Hercynian mountains created an orographic barrier between the ocean (Tethys) and NW Europe (Fig. 2), and would have enhanced the aridity expected because of the sub-tropical location. As the Triassic progressed, however, these mountains were eroded down, and the Tethyan Ocean encroached further west into Europe, bringing marine flooding to southern Britain in the Rhaetian. Both these effects would have ameliorated the climate.

It is widely accepted that monsoons were a regular feature of the Triassic (Crowley, 1994). This seasonality may have allowed moisture-bearing systems to penetrate into the otherwise arid regions, much as the summer monsoons bring irregular rainfall to central Australia today (Kotwicki & Allan, 1998). Comparisons with the countries surrounding the present-day Pacific suggest that superimposed on the annual variations in the weather systems of the Triassic would be longer-period fluctuations such as the El Niño-Southern Oscillation (ENSO) phenomenon (Kotwicki & Allan, 1998).

We must be careful about such predictions of the past, because many factors are non-uniformitarian. Prediction of Triassic dryland locations based on the known driving forces for Recent drylands must take into account unusual planetary events occurring around the Permian-Triassic boundary (see summary in Veevers et al., 1994). For example, coal-forming conditions during the Pennsylvanian in Euramerica, and Permian in the Gondwanaland province, were terminated by global warming that accompanied excessive venting of CO₂ into the atmosphere during the eruption of the Permian/Triassic (~250 Ma) Siberian Traps and other volcanics. The resulting gap in coal deposition lasted through the Early and Middle Triassic until the excessive CO₂ was finally resorbed by the crust (Veevers, 1994).

What we still struggle with is interpreting the subtleties of the climate variation in past times. The alternation between aeolian and fluvial and lacustrine litofacies, such as seen in the Lower Permian Rotliegend Group of NW Europe, implies significant and repeated climate fluctuation (e.g. see George & Berry, 1997). But we still have some way to go before we fully understand the driving forces, and the lithological successions produced.

BLAKEY, R.C., PETERSON, F. & KOCUREK, G. 1988. Synthesis of late Palaeozoic and Mesozoic eolian deposits of the Western Interior of the United States. Sedimentary Geology, 56, 3-126.

BEGIN, Z.B. & SCHUMM, S.A. 1984. Gradational thresholds and landform singularity - significance for Quaternary studies. Quaternary Research, 21, 267-274.

BULL, W.B. 1979. The threshold of critical power in streams. Geological Society of America Bulletin, 90, 453-464.

CROWLEY, T.J. 1994. Pangean climates. In: KLEIN, G.D. & BEACHAMP, B. (eds) Pangea: paleoclimate, tectonics, and sedimentation during accretion, zenith, and breakup of a supercontinent. Geological Society of America special paper 288, Boulder, Colorado, 25-39.

GEORGE, G.T. & BERRY, J.K. 1997. Permian (Upper Rotliegend) synsedimentary tectonics, basin development and palaeogeography of the southern North Sea. In: ZIEGLER, K., TURNER, P. & DAINES, S.R. (eds) Petroleum geology of the southern North Sea: future potential. Geological Society special publication 123, London, 31-61.

GROVE, A.T. 1977. The geography of semi-arid lands. Philosophical Transactions of the Royal Society of London, B-278, 457-475.

KOTWICKI, V. & ALLAN, R. 1998. La Niña de Australia - contemporary and palaeo-hydrology of Lake Eyre. Palaeogeography, Palaeoclimatology, Palaeoecology, 144, 265-280.

LEHOUEROU, H.N. 1968. La desertification du Sahara septentrional et des steppes limitrophes (Libye, Tunisi, Algerie). UNESCO, Paris.

MEIGS, P. 1952. Arid and semiarid climatic types of the world. In: Proceedings, eighth general assembly and seventeenth international congress, International Geographic Union. Washington, 135-138.

NANSON, G.C., CALLEN, R.A. & PRICE, D.M. 1998. Hydroclimatic interpretation of Quaternary shorelines on south Australian playas. Palaeogeography, Palaeoclimatology, Palaeoecology, 144, 281-305.

NANSON, G.C., PRICE, D.M. & SHORT, S.A. 1992. Wetting and drying of Australia over the past 300 ka. Geology, 20, 791-794.

OBERLANDER, T.M. 1994. Global deserts: a geomorphic comparison. In: ABRAHAMS, A.D. & PARSONS, A.J. (eds) Geomorphology of desert environments. Chapman & Hall, London, 13-35.

PATTON, P.C. & SCHUMM, S.A. 1981. Ephemeral stream processes: implications for studies of Quaternary valley fills. Quaternary Research, 15, 24-43.

PETIT-MAIRE, N. & GUO, Z.T. 1998. Mid-Holocene climatic change and Man in the present-day Sahara desert. In: ALSHARHAN, A.S., GLENNIE, K.W., WHITTLE, G.L. & KENDALL, C.G.St.C. (eds) Quaternary deserts and climatic change. A.A.Balkema, Rotterdam, 351-356.

SCHUMM, S.A. & HADLEY, R.F. 1957. Arroyos and the semiarid cycle of erosion. American Journal of Science, 255, 161-174.

SCHUMM, S.A. & PARKER, R.S. 1974. Implications of complex response of drainage systems for Quaternary alluvial stratigraphy. Nature, 243, 99-100.

SCOTESE, C.R. 1994. Early Triassic palaeogeographic map. In: KLEIN, G.D. & BEACHAMP, B. (eds) Pangea: paleoclimate, tectonics, and sedimentation during accretion, zenith, and breakup of a supercontinent. Geological Society of America special paper 288, Boulder, Colorado.

UNEP. 1992. World atlas of desertification. Edward Arnold, Sevenoaks, UK.

VEEVERS, J.J. 1994. Pangea: evolution of a supercontinent and its consequences for Earth's paleoclimate and sedimentary environments. In: KLEIN, G.D. & BEACHAMP, B. (eds) Pangea: paleoclimate, tectonics, and sedimentation during accretion, zenith, and breakup of a supercontinent. Geological Society of America special paper 288, Boulder, Colorado, 13-23.

WILLIAMS, M.A.J., DUNKERLEY, D.L., DE DECKKER, P., KERSHAW, A.P. & CHAPPELL, J. 1993. Quaternary environments. 2nd edition. Arnold, London.