HYPERPYCNAL TURBIDITY CURRENTS AT THE HEA D OF THE VAR CANYON - HYDROLOGICAL DATA AND GEOLOGICAL OBSERVATIONS

The Var River is 120 km long upon entry into the western Mediterranean Sea and drains a 2820 km(2) basin (Fig. 1). A steep submarine canyon connects directly to the river mouth (Fig. 2). The submarine canyon is sinuous and shows typical badland features such as high slopes result ing from erosion (Fig. 3). The average water discharge 152-53 m(3) s(- 1); Fig. 4) can be multiplied by tens during spring or fall ''flash fl oods'', when suspended sediment concentration can reach many tells of kg m(-3). The rating coefficient b corresponding to instantaneous disc harges is 1.534 (Eq. (1) and Fig. 5, curve 1), according to data publi shed by Laurent (1971), but might be larger if suspended sediment conc entration related to floods occurring after a dry period (Fig. 5, curv e 2) is taken into account. Nevertheless, concentrations predicted by (2) using a larger value of b (1.65 b less than or equal to 1.7) are n ot consistent with data published by Laurent (1971). For this reason, we used relationship (1) in the paper that follows, and our results ca n thus be considered as the minimum estimate of hyperpycnal plume prod uction. A definitive choice between relationships (1) and (2) in Figur e 5 requires additional measurements of suspended sediment lend, espec ially during peak flood discharges. The b value corresponding to month ly discharges is estimated to be 1.7 to 1.75 (which leads to mean annu al suspended sediment concentrations in the range 0.68-0.83 kg m(-3) R elationship (1) suggests that the critical discharge needed to produce a turbid hyperpycnal plume during floods is of the order of 1250 m(3) s(-1), depending on the salinity and temperature of the sea water nea r the river mouth (Fig. 6). A hyperpycnal plume is a turbidity current generated at a river mouth when thedensity of the river plume exceeds the density of ambient sea water due to sediment in suspension. The p lume plunges and becomes an auto-maintained turbidity current. The thr eshold is between 620 and 750 m(3) s(-1) using relationship (2) in Fig ure (5). Using relationship (1), short duration (<1 day) hyperpycnal p lumes may be produced every four years (Fig. 7a) and hyperpycnal plume s lasting more than one day may be produced every 21 years (Fig. 7b). Using curve 2 in Figure 5, the return periods are 2 and 5 years for sh ort duration and 1-day-long plumes, respectively. The Var river is thu s a river that can produce relatively frequent hyperpycnal plumes. as might be inferred from the study of Mulder and Syvitski (1995), who sh owed that a majority of rivers with low-to-medium average discharge (< 460 m(3) s(-1)) can produce such plumes. The calculated return period of these events at the mouth of the Var river could be greater at tim es of active delta construction, e.g. during ice ages, if no important change in precipitation occurred (Mulder and Syvitski, 1996). The Nov ember 1994 flood (Fig. 8) could have produced a hyperpycnal event almo st 20 hours in duration, and could have carried 18 x 10(6) t of sedime nt towards the deep basin. This represents 11-14 times the total avera ge yearly suspended sediment load (1.32 x 10(6) to 1.63 x 10(6) t for b values related to monthly discharges = 1.7 and 1.75, respectively). It is also the range of volume of sediment estimated by Habib (1994) i n connection with the failure at Nice Airport in 1979 (8 x 10(6) m(3)) . The November 1994 plume could have deposited at least 12 x 10(6) m(3 ) of sediment, irrespective of any erosion that might have occurred al ong the travel path. This should correspond to a turbidite layer many millimetres to many centimetres thick in the upper and median fan vall ey. We calculate center line values of the sedimentation rate in the V ar deep sea fan using the ID model for deposition from a spreading plu me at a river mouth published by Syvitski and Lewis (1992; Eq. (2) and (3)) using a width at the river mouth. w(o) = 250 m. Values are circu lated for a total suspended toad = 1.32 x 10(6) t yr(-1) (values into brackets provide values for a total suspended load = 1.63 x 10(6) t yr (-1) as a comparison), We use five removal coefficients (lambda) corre sponding to five grain sizes (from clay to fine sand) and an average l ambda = 6.8 day(-1) (medium slit) to describe the whole sediment. Resu lts show that (Fig. 9a), the sediment inventory at the river mouth var ies from 0.03 (0.04) to 0.43 (0.53) kg m(-2) d(-1) for clay and 0.12 ( 0.14) to 1.65 (2.O6) kg m(-2) d(-1) for fine sand in August and May (l owest and highest average monthly discharge, respectively). Using a di fferent removal rate for every grain size (Fig. 9b, see Bursik, 1995). fine sand cannot be deposited further than 13 kin from the river mout h, whereas clay can be deposited up to several tens of kilometres. The corresponding mean monthly center line values of theoretical accumula tion rates at the river mouth using a mean grain size (lambda = 6.8 d( -1)) (Fig. 10b) are 140 (175.2) cm yr(-1) in May and 10 (11.8) cm yr(- 1) in August for an initial flow thickness, H-o = 0.3 in August and 0. 58 m in May. For a total suspended load = 1.32 x 10(6), the mean annua l value varies from 48.7 to 99.7 cm yr(-1) depending on H-o. The mean annual center line sedimentation rate is 53.8 (66) cm yr(-1) for H-o = 0.3-0.58 m (Fig. 10). Using 5 grain size . 10a) provides identical se dimentation rate values at river mouth but larger values with distance from the river mouth because of the use of a constant men monthly gra in size distribution (Fig. 10a). Note that the center line values calc ulated using lambda = 6.8 d(-1) (Fig. 10b), can be considered to be mo re calistic but are still slightly underestimated close to the river m oth and slightly overestimated in more distal areas for two reasons: ( i) the peak of sediment concentration that occurs just before the peak of discharge during the rising limb of the flood hydrogram could not be measured (Laurent, 1971); and (ii) during high-discharge periods or igin and size of particles carried to the river mouth during mean-disc harge periods and tend to settle at a different distance from the rive r mouth (see Fig. 9). For example, using an average gram size for sedi mented particles corresponding to coarse silt (lambda = 12.3 d(-1)), t he center line value of sedimentation rate at river mouth is 88.8 (109 .1) cm yr(-1). Sedimentation rate values are lower using a model of a semi-circular plume in which sediment rate decreases as a square funct ion of the distance measured from the river mouth Eq. (4) an