The Water Crisis in Yemen

As described in Chapters 1 and 4, water has always been scarce in Yemen. Parts of the highlands do enjoy better rainfall than elsewhere in the Arabian Peninsula, but in Sana’a the rainfall would be barely enough to grow even a crop of sorghum were it not for the hand of man. Since prehistoric times, Yemenis have been adept at multiplying up rainfall by water management technology and careful husbandry. In classical times, Yemen’s irrigation civilization was renowned. Elaborate terraces and water harvesting structures, and adept management of springs and flood flows have allowed Yemen to support a large population, frequently 100–200 persons/km², one of the densest rural populations in the Middle East, and from time to time to grow rich on the export of products prized in the outside world – frankincense, myrrh, indigo, and coffee (Chapter 2). It was water management, more than water resources endowment, that earned Yemen the name ‘Blessed Arabia’.

Practices in the 1960s were recognizably the same as those described in mediaeval almanacs, and water structures were similar to those uncovered by archaeology from three millennia ago. The management of springs, spate flows and watersheds was a local affair with evolved systems of rights and responsibilities. Disputes were usually settled locally under customary law.¹ It is this heritage of traditional water management that kept Yemen green and supported its large population.

Rainfall in the desert margins is pitifully small and only irrigation could provide the food the ancient city states required. There is archaeological evidence of irrigation at Ma’rib in the late third millennium BC. Irrigation techniques and structures were progressively developed until, at length, a technology for controlling and diverting the massive spate flood flows was perfected.

Saba, the largest state, also had the largest irrigated area. A flood diversion and irrigation scheme was constructed in Wada Adhana, at Ma’rib, in the sixth century BC. The ‘Great Dam’ was a massive masonry structure, half a mile long (680 m), up to 60 feet high (18 m), and at its base more than 300 feet wide (100 m). The structure was not technically a dam but a weir, to stem the flow of flood waters and to raise the water level, so that the flow would run into two large intake structures. These can still be seen today, one on the north bank and the other on the south bank.

From the intakes, the flood flows entered stilling basins, designed to slow down the speed of the water and to allow some of the sediment to decant. Here, spillways threw water in excess of what the canals could carry back into the wadi bed. From the basin, the water flowed by gravity down a main canal on either bank to secondary canals. Drop structures along the way slowed the velocity of the flow and again spilled excess water from the canals back into the wadi. These techniques of diverting and slowing spate torrents, developed by the Sabaeans 2,500 years ago, are essentially unchanged to this day.

The works irrigated a ‘command area’ of 9,600 ha, although the area typically irrigated in any one season would have depended on the volume of water available. It has been estimated that 5,000–6,000 ha would have been irrigated and cultivated in a typical year.²

The system was several times damaged and rebuilt. Inscriptions record frequent breaches, particularly in the declining years of Saba when human and financial resources to maintain and repair the system were dwindling. Despite this, the system survived until the beginning of the seventh century AD, a total of 1,200 years, possibly the longest serving irrigation scheme on record.³

Irrigation and agricultural practices at Ma’rib closely resemble those used in spate irrigation today

Fields were rectangular with low earth bunds to allow the water to pond and infiltrate. Land preparation was by shallow ploughing, using a plough very similar to the wooden hook plough still in use in Yemen today. Although the primary source of nutrients was silt brought by the flood, Hehmeyer found evidence both of ash, probably from stubble burning, and of the incorporation of manure. Irrigation was through flooding, with water ponded in the fields between 60 and 80 mm. Annual crops were grown in two seasons, principally cereals: inscriptions mention wheat, barley and sorghum, and also pulses and vegetables. Two seasons were practised, corresponding to the two rainy seasons. Supplementary irrigation from wells and cisterns was used for tree crops, notably date palms⁴ but also vines and orchard fruits. Annual crops were grown around these trees, an inter-cropping practice not uncommon today, especially when trees are young.⁵

Cooperation over irrigation management in ancient Saba built social capital and helped to form state institutions

The farmers of ancient Saba cooperated to distribute water equitably to the fields, regardless of distance from the source. Regular communal maintenance of canals also built social capital in the form of water management institutions. These patterns of cooperation and institutional development seem to have been at the very origins of state formation in Saba in the eighth century BC.⁶

Inscriptions tell us a good deal about the organization of irrigation. When the spate arrived, the entire farming population would gather in the fields and an official would oversee the distribution of the flood waters. Allocation was on a time basis, with the hours calculated during the day by the length of the shadow of a staff, and at night by the movement of the stars. Water was managed by groups, usually family clans, and could be traded. An inscription records two clans turning over their quota to the water master to give to others, no doubt to be swapped for another turn at a later date.⁷

Major aspects of agriculture and irrigation were regulated by the rulers and by tribal assemblies. A king of the nearby Qataban state in the third century BC banned the sinking of new agricultural wells and restricted the use of water on certain communal lands. Another decree called for a new irrigation canal to cross the territory of another tribe. Stone tablets recording these decisions were set up on site.⁸

Dedications exist on tablets for well digging projects, for the repair of hydraulic structures, for the replanting of a palm grove, and for the levelling of fields to improve water distribution. One pleasant text describes the delivery of 300 camel-loads of wine for workers repairing the dam at Ma’rib.⁹

As with other Middle Eastern civilizations, Sabaean power and success depended on its irrigation systems, and the state invested in their building, maintenance and periodic reconstruction

Some 30,000 workers were employed at times for major maintenance. When flood damage carried structures away or badly damaged them, they were immediately rebuilt. The survival of the state depended on it. When the final destruction took place, Saba had lost both its political independence (to Himyar) and, with the dwindling of the incense trade, the sources of its wealth. The resources to maintain or rebuild the scheme were lacking, and the collapse of the independent polity and of its commerce and its source of food combined together to spell the end for Saba.

The final collapse of the structures and the destruction by flood of the irrigation perimeter at Ma’rib sent echoes throughout the known world

The collapse after 2,000 years of the political and economic power of the old South Arabian kingdoms was simultaneous with the collapse of the Great Dam of Ma’rib, so that the sweeping away of the hydraulic structures was seized on in imagination as symbolic of the sweeping away of an ancient civilization – and of the arrival of Islam.¹⁰ As Dresch says, ‘the collapse of the Ma’rib dam is the mythical moment of Yemen’s own collapse’.¹¹ The Qur’an sura Saba sets out the logic:

A sign there was to Saba, in their dwelling places […] two gardens, the one on the right hand, and the other on the left. But they turned aside: so we sent upon them the flood of Iram: and we changed their gardens into two gardens of bitter fruit and tamarisk, and some few jujube trees. (sura 34: vv15/16)

The chief divinity throughout ancient Southern Arabia, ‘Athtar, was god of rain.¹² ‘Athtar is associated with the ritual hunt that was conducted for the purpose of assuring rain. The Sabaeans, dependent for their power and survival on irrigated agriculture, traced their descent from their patron divinity, the god Almaqah, who was god of agriculture and irrigation. As in many cultures, magic and rain are often associated.¹³ Each year at Saba huge assemblies attended rain ceremonies in the vast temple of Awwam, dedicated to Almaqah, and from there they set off on their great annual pilgrimage.¹⁴ The Qur’an traces the very origins of life to water:

No literature survives from Saba. The vast quantities of tablets are all votive, administrative or commercial in nature. But we know from elsewhere in the peninsula that Arabian tribal society in pre-Islamic times had an oral literature of complex odes (qasida) of great vigour and imaginative and precise imagery. Water is a constant theme in this poetry. In fact, Nicholson characterizes the earliest Arabian poetry as ‘fountain songs, war songs and hymns to idols’. Those who worked with water had a special status, equal to poets and prophets. A youth was refused the hand of his beloved on the ground that he was ‘neither a poet nor a soothsayer nor a water diviner’.¹⁵ Rain is a constant metaphor for readiness and generosity:

So we are as the water of the rain shower: in our metal is no bluntness, neither is any miser numbered amongst us.

Irrigation may have developed as an early response to climate change five millennia ago

The hydraulic works of the city states were not the only examples of early water management. Throughout Yemen, from about 3000 BC, the monsoon appears to have weakened and the drying climate created incentives to maximize the amount of rainfall beneficially captured for agriculture. These incentives were sharpened by a growing population and the advent of new high-energy cereal crops.¹⁷

Spate irrigation was practised since prehistoric times not only by the city states but in all the wadis of south-west Arabia. Archaeological surveys in Wadi Dura, south-east of Ma’rib, for example, show evidence of spate diversion works from about 1500 BC onwards.¹⁸

Traditional spate technology was well adapted to the uncertain nature of spate floods

Over the centuries, farmers in the vicinity of wadis developed simple earth-built systems and irrigation networks to divert spate floods onto fields. Temporary dikes (ogma) were built across the flow channel using wadi bed material in order to divert flood water down canals or direct to fields. Often, successive dikes were constructed down the course of the wadi, each commanding a separate farmed area. When the upstream dike was breached, the spate flowed to the next dike and irrigated the fields there, and then in turn to dikes further downstream. This system of successive earth dikes respected the principle of ‘upstream first’ but the weak construction of the dikes ensured that when spate flows were abundant, water flowed in turn to downstream riparians, making for equitable distribution. The ponding of the water behind the embankment also helped infiltration to the underlying alluvial aquifers from where it was easy to lift the water out later in the season through shallow wells.

Farmers were responsible for organizing operation and maintenance, and for allocating the water. Typically water was allocated on the basis of the highest first, with rules about how much water each farmer might take before passing the water to his downstream neighbour. In most systems, the water channelled off at the diversion structure flowed down from field to field, and was released from the higher field when that field had ‘drunk’ – this may have been when the water had filled the field up to the height of a man’s knee or higher.¹⁹ In subsequent floods, the round might start with the last field left unwatered previously, as in Wadi Zabid (see Box 5.1). Further diversion structures downstream might receive water only when the upstream structure was breached, which in the case of traditional structures is a regular occurrence. Alternatively, there might be rules that a proportion of each flood must pass down or that the water in different periods was allocated to downstream diversion structures (as at Wadi Zabid, again see Box 5.1). In Wadi al-Jawf, the spring spate was allocated to one bank and the summer spate to the other.²⁰

Spate irrigation was the object of both state and local community development efforts

Where there was a sufficient degree of organization and capital, larger spate wadis were developed systematically with masonry strictures. The Cyclopean works at Ma’rib, the stone-built dikes of the Himyaritic era, and the sophisticated hydraulic structures and diversion regimes in Wadi Zabid under the Rasulids, are examples of these more elaborate systems. The masonry systems had the advantage of being more permanent and more efficient in diverting the flows. The disadvantage was that upstreamers could monopolize the water, as the masonry structure was rarely breached. Rules to ensure fair shares for downstreamers were thus often developed to accompany the development of these more permanent structures, as in the case of Wadi Zabid discussed in the box above. The smaller wadis in the mountains were developed over the centuries by farmer groups.²²

Box 5.1: Historical spate management in Wadi Zabid.²¹

Water in Wadi Zabid’s spate system is managed according to rules codified in the 1390s by the qadi Muwaffaq al-Din al-Nashiri following violent disputes among farmers. Later these rules were used by the Moslem scholar Sheikh Ismail al-Jabarti to arbitrate disputes.

The rules modified the allocation of spate water according to the shari’a principle of upstream first – al ‘ala fa al ‘ala – by dividing the waters between three ‘regions’ by calendar dates. These rules are essentially still in operation today. In the dry season, there are baseflows only, which go to the upper region. In the rainy season when the floods come, the spate is divided between:

• The upper region for the first 97 days.

• The middle region for the next 20 days.

• The lower region for the next 35 days.

The channel master allocates water to plots and decides which plot will get irrigated first by the next flow: the rule is ‘not twice in 14 days’. The channel master collects charges proportional to the irrigated area: the rate is lower in the middle area, and negligible in the lower area.

Despite these rules, many conflicts over water distribution are recorded, which in the past are said to have claimed several lives each year.

Tanks and cisterns (Figure 5.1) have a very ancient pedigree and exist throughout the highlands and in other locations suitable for capturing rainfall and run-off. The 17 consecutive Tawila tanks at Aden, cut into the rock and built up in stone, date from the Himyaritic period. They probably supplied drinking water for the city, which was again their use after their rehabilitation by the British in 1856.²³ Smaller systems, often of ingenious construction and great antiquity, can still be found throughout Yemen. They were used either for local drinking water or for watering stock. Generally rectangular, there is usually a flight of steps for access. Those used for drinking water were generally covered, some of them dug underground and carefully lined with stone. Run-off slopes from which the rain water was ‘harvested’ to flow into the tanks were scrubbed and kept scrupulously clean of litter.²⁴

Figure 5.1 A traditional village water cistern (Shihara). Photograph courtesy of Peer Gatter.

Springs have more stable flow regimes than spate water, and hence the quantity of water to which each riparian is entitled is more predictable. Water rights were normally attached to the land and denominated in time shares – so many hours once a fortnight, for example. In Wadi Dahr, rights to spring water were divided exactly between lower and upper communities, 15 days each. These block allocations were then further divided within each community according to time-share ownership rights. These rights could be exchanged with other users to suit the farming need. Water distribution was overseen by local irrigation supervisors.²⁵

Harvesting of run-off was an important source of water in the inter-montane plains

Run-off rights were assigned from specific slopes to parcels of bottomland in a proportion – sometimes up to 20 times the area of bottomland – adequate to grow a crop on the run-on land. These run-off rights were so important that it is claimed that, historically in Sa’adah, houses had no cisterns for domestic water as this would infringe the run-off rights of others.²⁶

These terraces sometimes descend precipitous slopes in a long series (Figure 5.2). The terraces, many of enormous antiquity, collect and retain the run-off. They are faced with stone, which helps retention of soil and water. It is estimated that, using traditional tools, a family work party of a few men could construct a terraced field of half a hectare in about a month of labour.²⁷ The cultivable surface is sometimes tiny – no more than a few square metres. These terraces also play an environmental function in reducing floods and landslides and preventing erosion.

The earliest evidence of terraces has been provided by archaeological exploration in the Khawlan area near to Dhamar, south of Sana’a, where the predecessors of present-day terraces date from the Bronze Age, as early as the third millennium BC. Soil analysis suggests that in some areas the slopes had been forested, and that erosion had followed the destruction of the forest cover in the pre-Bronze Age period (perhaps 5000–7000 BC). Terracing thus responded to the need to protect the slopes while expanding the arable area. Evidence from these Bronze Age sites shows that barley and wheat were grown, together with legumes such as lentils and common peas, and perhaps chickpeas.²⁸

Figure 5.2 In this remote village, ancient terraces are still cultivated with cereals – but also nowadays with qat (Wadi Qadaha, Al-Mahweet Governorate). Photograph courtesy of Peer Gatter.

The kingdom of Himyar, which flourished from the second to the sixth century AD, was characterized by intensive agricultural terracing. The valleys around the Himyarite capital of Zafar are almost entirely stabilized by a complex system of dams and huge terrace walls. It is likely that much of the characteristic terraced highland landscape was created at this period.²⁹

The archaeological record also demonstrates the life cycle of a terrace system

In Wadi Shalalah, the old Himyaritic irrigation system disintegrated after the fall of Himyar. Local people then constructed terraces in the side valleys between the twelfth and seventeenth centuries AD. However, in the twentieth century erosion from the top of the mountain, perhaps associated with overgrazing, led to catastrophic landslip and the collapse of rock and gravel debris from above onto the terraces. Over a number of years, floods sheeting down the slope destroyed sections of the terrace retaining walls. The system has now collapsed.³⁰

Check dams and retention structures have been constructed in the highlands since earliest times

Gibson and Wilkinson found evidence from the Himyaritic period of up to 100 small dams near Dhamar.³¹ The dams are of two types, both well built from masonry. High dams were built across small valleys to impound the seasonal flows. Lower dams – essentially weirs – were built across broad valleys to raise the water level above the fields and to encourage infiltration into groundwater. Archaeological evidence shows that two of the high dams were breached in antiquity. The lower dams were not breached, which suggests that flood volumes and velocities were at some stage being moderated both by the higher retention dams and by the presence of terraces and check structures throughout the watershed. Effectively, the Himyarites developed a system of integrated water management that maximized water retention, reduced soil erosion, and checked and retained flood flows within a field system in the valley bottoms. The tenth century Yemeni historian al-Hamdani correctly attributed these old dam structures to the Himyarites, and modern villagers apparently do the same. Wilkinson found that villagers still call irrigation features by their old Himyaritic names.³²

When state power weakened, many dams were abandoned or integrated into terrace systems

Gibson and Wilkinson conclude that the dams, which required considerable capital and organization, were operated under central authority during the period of Himyaritic power but that after the collapse of the Himyaritic kingdom in the mid-sixth century AD, local people were unable to maintain them and they gradually went out of use. Again, as at Saba, this is testament to the links between state power and water development in the period (see section 5.2). Terrace agriculture, however, survived because it could be managed by small household-size units. In fact, the offtake sluices of many ancient dams have been sealed and the structures now serve as the retaining walls of sizable terraces.³³

Until recently, hand-dug wells were the principal means of extracting groundwater

In historical times, every landowner had the right to abstract groundwater by digging under his farm. There were some traditional rules over access – e.g., harim well-spacing rules, which might forbid a new well within 500 m of an existing one – but no traditional rules over quantities extracted or water charges.³⁴ Traditional dug wells have diameters between 0.5 m and 1.5 m, and were dug as deep as 60 m. Wells were dug by hand by professional well diggers, and were unlined or lined with stones. In the past, well digging was a significant event in the community. Serjeant describes the traditional digging of a well in the Hadramawt: before commencing, a sheep was slaughtered, again when water was struck, and a third beast was sacrificed when the lining was complete.³⁵

The typical method of drawing water from a hand-dug well was for a man to run himself or to drive a donkey down a long catwalk. The bucket rose up and tipped mechanically into a waiting channel or receptacle. From about the 1930s, mechanical pumps powered by donkey engines were introduced. Although ‘modern’, these motorized wells, now largely replaced by tubewells, already have the air of heritage about them where they survive. Their gentle ‘phut phut’ is still a pleasant sound on a hot afternoon in a dry wadi (Figure 5.3).

As early as the twelfth century AD, problems of falling water table were experienced

Over-extraction of groundwater was in the past not much of a problem: with only shallow wells and manual extraction, the quantity of water that can be drawn is limited. Nonetheless, Croken records that between 1183 and 1229 AD, the water level in wells in Zabid dropped by 30 m. This early example of groundwater depletion was probably the result of a large increase in the population of the city after the Ayyubids made it their capital.³⁶

In various parts of the country, systems for conveying water from underground springs were used (Box 5.2), on the same principles as the Iranian qanat and the falaj found in Oman. The first evidence of these structures in Yemen dates from the fifth century BC.³⁷ Underground tunnels were typically hand-dug over sometimes long distances, up to 25 km. They can easily be spotted from the air, as every 50 to 100 m or so there is a shaft used during construction and surrounded by the excavated spoil. Once the tunnel was excavated, the shafts were used for maintenance.

Figure 5.3 A hand-dug well in the Hadramawt. The gentle ’phut-phut’ of the donkey engine is a pleasant sound on a hot afternoon. Photograph courtesy of Matthias Grueninger.

Box 5.2: An ancient qanat system near Sana’a.³⁸

Dhelaa, near Sana’a, has a long history of irrigated agriculture. The area receives surface water from the 4 km Matba tributary of the Wadi Dhelaa and in the past was supplied both by surface flows from the river bed and by ancient qanat that brought water underground from the nearby mountains. Inscriptions inside the tunnel of the qanat suggest that they were developed at least 2,000 years ago. Over the years the tunnel was gradually deepened to keep up with fluctuating water tables.

Unlike the Iranian qanat, which are dug through the compacted sediments of an alluvial fan, in Yemen the channels were often cut through rock. The techniques were long familiar to Yemenis: many of the channels and structures at Ma’rib were cut through rock, and at Baynun the Himyarites carved out a tunnel to carry water right through the mountain.³⁹

One such system was constructed in the karst formations in the limestone at Ghayl Ba Wazir near Mukalla, where fissures contain large volumes of fresh water. Local legend records that Sheikh Umar Ba Wazir initiated the digging of the gently graded tunnels that linked the fresh water deposits to the town. A dense network of underground channels criss-crossed the area. Typically, the first use was domestic. Where the channel passed a settlement, stairs were constructed to allow water to be drawn for domestic use. There were also public laundry points, and facilities for bathing. The channel then served irrigated gardens and orchards. Water could be switched between outlets according to water rights, which were apportioned by time. To ensure that these time shares were respected, a star calendar told the night-time hours, and the sun’s shadow was used by day.⁴⁰

Nowadays, the drop in water levels at Ghayl Ba Wazir has led the ground to cave in, revealing vast holes in which the ancient tunnels which once bore the water can be seen, high and dry, well above the present water level.

This elaborate system tapped the groundwater and springs of the Hadda area, to the south and above the city. The system is very old, first documented in 799 AD as constructed by the Abbasid governor of the city, but its origins are probably much older still.⁴¹ Three long conveyors, sometimes underground, sometimes in an open canal, carried water some 10 km to the city and for agriculture in the Sana’a plain. Extensions ran beyond Sana’a to bring water to al Rawdah, beside the present airport. These conveyors were given names, sometimes rather glorious ones like al Mahdi (the rightly guided), or al Mansur (the victor) or al Aswad (the swart). Within the city, the water was distributed to the mosques in each quarter, for drinking and ritual ablution. The sullage water then irrigated the neighbourhood gardens that exist in the Old City of Sana’a to this day.

The system provided water for the entire population and for thriving market gardens

It is estimated that the flow in the Sana’a ghayl was several hundred litres per second, a considerable volume equal to more than 25 million litres a day, enough to provide domestic water for all the population – probably about 50,000 people – and to irrigate vegetable gardens and vineyards.⁴²

Croken describes the construction some time before 1167 AD of a similar underground conveyor from springs to the city of Zabid. It was constructed by Qadi al Rashid abu’l Hasan, ‘the author and poet from Aswan […] the first of his age in engineering and religious sciences’.⁴³ Recent archaeological work may have uncovered part of this system, which was used for urban water supply and to irrigate gardens in and around the city.⁴⁴ Messick gives a fascinating description of the mediaeval water supply of the town of Ibb, which was carried through saqiyya (water channels) from the mountains to the east of the town.⁴⁵