Silting up of harbours

The ancients often looked for estuaries to shelter from the sea (and also to find fresh water). In this way they solved the problem of exposure to waves, but fell into another problem: the silting up of harbours by fluvial sediment (case of Ephesus and many other ports like Leptis Magna, etc.).

Sediment brought by rivers is usually transported by waves along the coastline on both sides of the estuary (this is called littoral drift). The direction and volume of this littoral drift is determined by the angle of incidence of waves arriving on the coastline. If a port is built in such an area, sedimentation must be expected on one side of the port with erosion of an equal volume on the other side (Portus Claudius, Caesarea Maritima).

A partial opening of the breakwater (e.g. arched breakwater at Puteoli) does not change much to the problem of silting up as the actuator of littoral drift is wave action. But the purpose of a breakwater is exactly to protect from wave action; hence, sand will settle down. However, a canal through the breakwater at the average wave-breaking line where a current is generated by wave set-up may help to flush the port basin (e.g. El Hanieh (Libya), Centumcellae (Italy), Caesarea Maritima (Israel), Sidon (Lebanon)).

Oblique waves generate an oblique coastline that tends to be oriented parallel to the wave crests, e.g. a tombolo is created behind an obstacle because of wave diffraction, like at the Giens, and Argentario-Orbetello peninsulas. Ancient places like Tyr, Pharos, Peniscola (Spain) and Peniche (Portugal) are also the result of large-scale tombolo development.

Empories’ ancient city wall acting as a detached breakwater (photo: A. de Graauw, 2008)


This is also true for so-called “detached breakwaters” (located at a small distance and parallel to the coast line). Sooner or later, a tombolo will develop behind such structures.


This problem of littoral drift is still encountered by modern coastal engineers on almost every coastal project. Let’s see this in more detail.

Littoral drift is quantified by several more or less complex formulae. We mention here the most popular and simple one, as proposed by CERC in 1984 (American Coastal Engineering Research Center):

where Q is the littoral drift (in m3/year), K is a coefficient (depending on parameters like wave steepness, sand grain sizes, etc.), H is the wave height at breaking (in m) and θ is the angle of incidence of waves on the coastline at the breaker line (in degrees). This formula shows the importance of the wave height, as anyone would suspect. It also shows the importance of wave incidence: littoral drift is nil with frontal waves (when wave crests are parallel to the coastline, θ = 0°), it increases with wave incidence up to 45° and reduces beyond that. The average wave direction thus determines the volume of sediment transported along the coastline and a sound knowledge of the wave climate and of wave propagation to the coast is required.

The main difficulty of computation of the coastline evolution is that waves reshape the sandy sea bed. This leads to an “iterative” computation of wave refraction and diffraction: the larger the wave incidence, the larger the littoral drift and the more the sea bed is reshaped, which in turn changes the wave propagation pattern and requires a new computation, etc.

Without going into further details, it can be understood that river sediment supply will be distributed on both sides of the estuary, generating a curved coastline in order to reduce the wave incidence with increasing distance from the estuary (a nice example is Ostia). It can also be understood that oblique waves generate an oblique coastline that tends to be oriented parallel to the wave crests (e.g. a tombolo is created behind an obstacle, such as the Argentario and Torre Vendicari peninsulas).

Similarly, for a bay between two rocky promontories: the shape of the bay will be curved (close to logarithmic spiral) corresponding to wave spreading due to refraction on the sea bed and to diffraction around the promontories (e.g. bays of Cavalaire, Pampelone, Alexandria’s Magnus Portus and so many others).

For wave incidence larger than 45° with respect to the coastline, a sand spit develops, e.g. Flèche de La Gracieuse near Fos where the modern port of Marseille has located its largest container and oil terminals. The sand spit usually ends with a hook due to wave diffraction. Sometimes, successive hooks can be seen as a result of long term evolution. A sand spit is often very narrow (say 20-50 m) and much effort is devoted to avoid its break-through during storms if it protects major infrastructures like at Fos. This author suggests a similar sand spit may have protected the entrance of Marius’ canal.

Our aim is not to summarise here one year of hydraulic courses for coastal engineers within one page, but

to stress the importance of wave action
and to note that this knowledge is available since only a few decades.