The tsunamis that devastated coastal areas in the Indian Ocean in 2004 and in Japan in 2010 raised global awareness of tsunamis, prompting government ministries to seek more information about the phenomenon. The Caesarea tsunami study began in 2006 (Reinhardt et al. 2006; Goodman-Tchernov et al. 2009) with the aim of documenting historical tsunami events and better understanding possible tsunami scenarios along Israel’s coastline. In the first years of the study, cores were drilled in the seabed at several locations off the port of Caesarea. The cores were analyzed and found to contain layers of sediments and shell fragments sandwiched between layers of Nilotic sand. The layers were dated by pottery and OSL analysis, and two strata were attributed to historically documented tsunami events: one on 13 December 115 CE, and the other on 9 July 551 CE (Goodman-Tchernov et al. 2009). The object of the 2010 excavations was to see if the stratification detected in the cores could be identified again in a marine excavation near Caesarea’s port and in a land excavation.
Following a storm in 2010, a section was exposed inside a Byzantine structure that stood on the edge of the cliff south of the Aqueduct Beach (Fig. 1: A). Subsequently, two excavations were conducted—a land excavation of the section revealed by the storm and an underwater probe, c. 850 m off the shoreline and 10 m below sea level, at a point almost directly west of the structure (Fig. 1: B)—with the aim of comparing their stratigraphic sequences.
The Land Excavation. The west-facing section was prepared for documentation with minimal archaeological intervention; this consisted mainly of straightening and lightly cleaning the section for photography and measurement purposes (Figs. 2, 3).
A layer of sterile sand (L121) was found at the base of the section. The archaeological strata cut into a layer of beach sand and mollusks that was devoid of finds (L117). Above the sand layer was a floor (L120) overlain by a layer of sand and potsherds (L118, L119; not drawn). A potsherd fill (L116; not drawn) overlaid the sand and pottery layer, and a drainage channel (L115) made of stone sections was placed on top of the fill. A layer of gray sand and potsherds (L111; not drawn) was found beside the channel. Kurkar foundation stones (L110) placed above the drainage channel and the gray layer were overlain in turn by several layers of plaster (L108, L109), which served to level the surface prior to the laying of a plaster floor (L107). These were overlain by remains of a wall (W04), beneath a bedding of plaster and shell fragments (L104), above which was an industrial white mosaic pavement (L103; 8.36 m asl). Two walls (W1, W2) delimited the floor to the south and north. The floor was covered with a layer of dune sand (L102) beneath a surface layer that reached the cliff top (L101; 9.16 m asl). Cleaning the section yielded a bowl (Fig. 4:1), cooking pots (Fig. 4:2, 3) and jars (Fig. 4:4–7) from the Roman period (first–fourth centuries CE), as well as an amphora (Fig. 4:8) and jars (Fig. 4:9–12) from the Byzantine period (fourth–seventh centuries CE).
The Marine Excavation. The excavation was conducted inside a metal frame (2 × 2 m, frame depth 0.6 m, excavation depth 1.5 m; Fig. 5), developed by the Marine Archaeology Unit for stratigraphic excavations on the seabed. A sequence of various sedimentary layers was revealed (Fig. 6): sand (L107); broken shells with sharp edges (L106); sand (L105); mud (L104); sand (L103); beach rock, potsherds, marble fragments and shells (L102); and sand (L101). Layer 102 yielded pottery (not drawn) dated to the Byzantine period.
The data provided by the cleaning of the section on land attests to the presence of a settlement in the Roman and Byzantine periods, at which time the coastline lay farther west than it does today; however, the section contained no evidence that could be attributed to a tsunami. In contrast, the sequence of stratification from the marine excavation attests to two tsunami events. Evidence of the earlier of the two is provided by the layer of broken shells (L106) between two layers of sand (L105, L107). Evidence of the more recent event is visible in Stratum 102, which dates from the Byzantine period and is sandwiched between two layers of sand (L101, L103). The layer of mud (L104) may be related to another tsunami event, as tsunami sediments are sometimes identified by the presence of a ‘mudcap’ (Gelfenbaum and Jaffe 2003). Alternatively, the layer of mud may have been formed by sediments originating in a Byzantine drainage channel leading into the sea; an opening for such a channel was found 70–80 m to the south of the building. However, this theory can only be confirmed by conducting further research.