Greece is the Word
CU 1999 Contents Page Next:
Austria expedition archive

Cambridge Underground 1999 pp 21-25

CUCC Recce to Katavothra, Epirus, Greece


This report describes the findings of the ExCS reconaisance to Katavothra. The report has been lodged with the BCRA library and the Greek Spelæological Society, as well as the team members.

Members: Wookey, Tess Jones, Tony Rooke, Sam Lieberman, Tanya Savage, Dave Ramsay

Dates: 27th April 1997 to 3rd May 1997


CUCC were invited in 1996 by Patrick Harris, a British resident of Margariti, to investigate the Katavothra sink nearby. It was noted that a very large amount of water disappeared into this sink at the end of the winter wet season. There is also a known subsea resurgence at Ag. Kiriaki, 4 km to the SE of the sink.

We must thank the Geological Institute at Prevezza which provided us with a geological map of the area at 1:25000 and permission to investigate the sink and try to trace the water. Co-ordinates were recorded using a GPS (Geographical Positioning System) with the WGS84 spheroid.

The area that the sink drains is an alluvial plain of Quaternary deposits. This plain is covered by up to 3m depth of water for the winter months, over an area of approximately 2 square kilometres. (~2x109 litres). The strata in the area dip to the NE. The Katavothra sink is at the SE boundary of the plain where it meets limestones of the Palæocene-Eocene period.


We first examined the sink on 1997.04.28. A sizeable stream was seen to sink at the foot of a 20m high outcrop (N39° 20' 24.6", E020° 24' 52.4"). Investigation showed that direct entry to the sink would require removing some large boulders, made difficult by the amount of water ponding at that point. Examining the cliff face, several possible entrance sites were found that might provide a way in. One (A) was just above the main sink, and contained detritus showing that it took water at high water levels. There were two others (B & C) -- a pair of holes 8m higher up the cliff, one above the other. Moving a few rocks allowed access to the bottom entrance, which comprised 8m of rift passage paralleling the cliff face, and containing a significant population of cave crickets. The higher entrance was another route into the same rift, although the route between the two was blocked by collapse.

Entrance A, though smaller, proved more promising. Some digging deepened the entrance enough for tight access for 5m to a tight bedding-plane squeeze. Beyond this was 10m more awkward passage to a wide collapse chamber, which contained a (live!) fox on initial examination. To the south-east the chamber extended down-dip to a point where the Katavothra water could be clearly heard.

This was what we found on the first day's exploration. Further work was curtailed when it rained a great deal overnight, so that by the next morning the water level at the entrance had risen by nearly 2m, nearly up to our new entrance. With further rain threatening it was not safe to enter, and indeed the water rose to cover the entrance by that evening. We temporarily blocked the entrance to prevent more debris filling it if the water got too high, and then took measurements of the water level at various times - see Fig. 1 for an approximate response of the sink to rainfall.

We looked for other springs marked in the area, and some sinks described by locals to the NW but found neither.

We measured the flow rate (see Appendix 3) on 1997.04.30 at about midday. This was about 1.8 m³/s. By midday 1997.05.02 (2 days later) it had fallen to about 1.1 m³/s. Given that the water level was falling by the time of the first measurement, it suggests that the cave has a flow capacity of about 2.0 m³/s. This estimate is complicated somewhat by the hydrostatic pressure built up in the lagoon that forms above the sink at such times, but gives a reasonable approximation of the situation. It is possible that opening up the entrance could have a significant effect on this value, which might have implications for how fast the wetlands drain, or even if they will remain inundated during the winter at all. Conversely, the maximum rate may well not be set by the apparent partial blockage at the sink, but by constrictions further down the system, in which case engineering at this point should have no effect except to enable entry.

We eventually got back into the cave on the last day, and surveyed it, including a new section off the chamber to the NW which lead to a short section of streamway, sumped at either end. This streamway is not the major one sinking at the entrance, but a much smaller one draining from east to west. The total surveyed length of the cave is 86m, with a vertical range of 7m.

Our conclusion is that this remains a very interesting site, but significant digging work will be needed to follow the water into the cave, with some large boulders to be moved. Very low water flow will be needed for this, so sometime in the summer seems likely to be best. There must be a significant cave here, although it may be entirely submerged. It should also be noted that the rock in the cave was thinly bedded and somewhat friable in places. This may mean that any cave has high levels of collapse within it, or this could just be the effect of this explored cave being extremely close to the surface.

Again we thank those who made this exploration possible, particularly Patrick Harris, the Geological Institute, the local Police, and those who helped with translation, accommodation, equipment etc. Everyone made us feel very welcome.

Appendix 1 Noted features of SpelŠological interest

(GPS readings on WGS84)
Katavothra Cave: N39° 20' 24.6", E020° 24' 52.4"
Significant bedding-controlled cave, 86m long, with tight entrance, large, low chamber and short section of streamway. Surveyed CUCC 1997.05.
Site 1: N39° 17' 08.2", E020° 27' 51.1" or N39° 17' 02.8", E020° 27' 53.9"
4m long truncated section of very well decorated cave in road cutting. (On inside corner of Margariti to Parga road near point where you first see the sea and the Ag. Kiriaki subsea spring.
Site 2: N39° 18' 45.7", E020° 29' 52.3"
2m long hole in wall. In quarry on Kanaliki to Morfi Rd.
Other sites examined:
Cliff and accompanying scree slope along side of Paleokastro lake: N39° 23' 01.9", E020° 24' 01.5".

No caves were found in this area.

Appendix 2: Katavothra Sink Rain Response data

Note that after heavy rainfall the flow quickly increases to the point at which the sink starts backing up, so the water level rises rapidly. Peak ponding level was reached about 24 hours after the rainfall in this case, peak flow rate obviously sometime before that. The level then slowly declined over the next 2-3 days to get back to within a metre of the original level.

The rainfall level on this graph is relative and approximate, heavy rain overnight on Monday evening, with persistent light rain for the next couple of days.

DayTimeDataLevel (m)
SundayAll day3.1m below 'GPS'0
MondayA.M.3.1m below 'GPS'0
Tuesday10:000.5m below 'GPS'2.0
Tuesday14:300.1m below pt33.0
(Max level)?0.4m above pt33.5
Wednesday12:000-5m below 'GPS'2.0
Friday11:301.5m below 'GPS'1.0
Friday17:001.7m below 'GPS'0.8

Figure 2 --- Water level data points

Note that the 'max level' value was determined from the tidemark, so the time is not known accurately. There were two reference points. The rock tip upon which the GPS unit was originally placed 'GPS', and (when this was inaccessible, or underwater!) a scribed 'X' by the entrance, point 3 of the survey.

Appendix3: Flow rate measurements

We made crude measurements of the flow rate and stream cross section to get an idea of the volume of water. The flow rate was obtained by timing small twigs thrown into the central flow over a measured distance. These were timed with a stopwatch and the results averaged. This was done on two occasions, once on Wednesday, and once on Friday, both around noon. The section used was next to the depth post used as station 1 for the survey. The cross section changed noticeably as the water level in the stream dropped. Obviously these measurements are all rather approximate and the stick method and simplistic calculations will overestimate the total flow. Nevertheless it gives some idea of the flow, and the figures are all given here should anyone wish to make a better estimate.

1997.04.30: (Wednesday)

Times to float 13.5m: 13.9s, 14.3s, 14.9s Average: 14.4s

Cross section: Approx. area: 1.9m²
Velocity: 13.5m/14.4s = 0.94 m/s
Volume:0.94 x 1.9 = 1.8 m³/s

1997.05.02: (Friday)

Times to float 13.5m: 22.5s, 17.5s, 18.5s. Average: 19.5s

Cross section: Approx area: 1.1m²
Velocity: 13.5m/19.5s = 0.69 m/s
Volume:0.69m/s x 1.1m² = 0.8 m³/s

As the water level at the sink was already going down by the time of the first measurement, it seems that the sink can take about this flow rate. A little more than that and it starts backing up. Nearly 2 cumecs is a very sizeable flow, and suggests a significant cave beneath the sink.

Appendix - Maps and Surveys

The original survey data are not reproduced here, but are available in the published report, or from the author.

Greece is the Word
CU 1999 Contents Page Next:
Austria expedition archive