Lessons learned from the maintenance of ancient aqueducts through their carbonate archives

The study of ancient aqueducts shows us how past societies used water resources such as rivers, lakes, springs and groundwater. Such resources were tested for their quality, captured, sometimes diverted or abandoned, or supported by a new source. The efficient management of ancient water supply systems was adapted to societal needs, to the chosen environment and to change in climate and population dynamics, similar to factors shaping our modern water management. The in-situ carbonate deposits of ancient water supply systems reflect characteristics of the water resources and aquifers used, and therefore constitute an archive with considerable research potential, particularly for better understanding of ancient water management, palaeo-environment, and water scarcity. Ancient aqueducts were long-lived structures, with recorded operation periods of up to 800 years such as the more than 400 km long Valens aqueduct of Constantinople. However, carbonate sequences are often shorter because of cleaning. The Valens aqueduct has a double channel to allow cleaning without interruption of the water supply, and shows only 27 years of carbonate deposits. A carbonate sequence of 250 years is documented for the ancient city of Nemausus, which correlates with the actual documented years of operation. This is because some aqueducts were built wide enough to allow smooth operation without being cleaned of carbonate for most of their length. The aqueduct of Arelate was modified by reallocation of branches and expansion, probably in response to increasing and changing demand of water. In at least 50 years of carbonate stratigraphy, this aqueduct shows periodic variations in the aquifer. As the focus of this paper we discuss the Roman aqueduct of Divona, which shows evidence of regular cleaning, repairs, modifications and even timing of maintenance, captured in aqueduct carbonate. The d18O profile of the carbonate stratigraphy shows a cyclicity that is inferred to reflect annual water temperature variations in the channel. This cyclicity was used to determine the timing of maintenance. The aqueduct operated for centuries but the preserved profile shows 88 years of operation, interrupted by two plaster repair events, representing time breaks in water supply of at least several months. Not less than 14 carbonate removal events were recognized, which took place in fall, spring, or winter but never in summer, with an interval of 1 to 5 years. Carbonate removal from channels was done rapidly, in less than one month, since stable isotope cycles show no indication of a long-term interruption in the water supply. The last maintenance was done only three years before the final abandonment of the water supply, showing that continued operation of the aqueduct was planned but was not fulfilled. This micro-story shows how a historical water management operated and adapted to water scarcity and population growth and decline, similar to modern days. Frequent cleaning of aqueduct carbonate and repairs are examples of what we call today sustainable water management. Ancient aqueducts carbonates are models to study aspects of past water management and adaptations to crisis, and provide a vision for the future of water availability and climate change.