Where Does a River Come Out of a Cliff? Water With Its Own Plumbing

Limestone dissolves slowly in slightly acidic water, and rain becomes slightly acidic by absorbing carbon dioxide from the air and from soil. Water entering cracks in limestone therefore widens them, and a crack that widens carries more water, which widens it faster, so the system develops unevenly with a few routes capturing nearly all the flow while the rest stay tight. Over tens of thousands of years this produces an integrated network of conduits, effectively an underground river system with tributaries, that carries water from where it sinks to where it emerges. The emergence point is determined by where the network meets an impermeable layer or a valley floor, which is why a spring can appear partway up a cliff.

Almost every property is different from water moving through sand or gravel:

• •Flow is fast, measured in kilometres per day rather than metres per year
• •Discharge responds to rainfall within hours rather than over months
• •Water is barely filtered, since it moves through open conduits rather than pores
• •Bacteria and pollutants travel intact over long distances
• •The catchment feeding a spring may not match the surface catchment at all
• •Flow can vary by a factor of a hundred between dry and wet conditions

Working out where a spring's water comes from cannot be done by looking, so the standard method is to introduce a traceable substance at a sinking stream and watch for it downstream. Fluorescent dyes are used most, detectable at concentrations far below anything visible and distinguishable from each other so several routes can be tested at once, with detectors left in springs for weeks. The results are frequently surprising, showing water crossing beneath surface watersheds, travelling many kilometres in a day, or a single sink feeding two springs in different valleys. Those maps matter practically, because they determine which land surface affects which water supply, and protecting a spring requires knowing that a farm several valleys away drains into it.

The surface over such a system is as distinctive as the underground part and the features are diagnostic. Sinkholes form where the roof of a cavity collapses or where soil is washed down into a widening fissure, and they appear abruptly, occasionally under buildings. Streams flow across impermeable rock and then disappear entirely on reaching limestone, leaving a dry valley continuing downstream that only carries water in extreme conditions. Bare limestone pavement forms where soil has been stripped, with deep fissures separating flat blocks. Dry valleys, closed depressions and an absence of surface streams over wide areas are all consequences of drainage happening underground. Around a fifth of the ice-free land surface is this kind of terrain.

A very large share of the world's population drinks water from limestone aquifers, with estimates commonly around a quarter, and the properties described make that supply unusually vulnerable. Contamination reaching a sinking stream arrives at the spring within days, carrying bacteria that would have been filtered out in a sandy aquifer, which is why outbreaks traced to such supplies recur. Turbidity spikes after rain as sediment is flushed through the conduits, which disrupts treatment. Yield collapses in drought because there is little storage in the rock itself. And the catchment being invisible means protection zones drawn by surface topography can miss the land that actually matters, which tracing studies repeatedly demonstrate.

Acidic water widens cracks in limestone, and cracks that widen capture more flow, so an integrated conduit network develops and delivers water to a spring that can emerge partway up a cliff. Flow is fast, responds to rain within hours and is barely filtered. Dye tracing maps the connections, which often cross surface watersheds, and around a quarter of the world drinks from such systems.