In their estimations of sedimentary transit, studies based on regular sampling, or a fortiori based on the extrapolation of limited measurements with SM = f(Q) relationships, seldom highlight bias due to certain human activities directly affecting the watercourse (maintenance, consolidation of riverbanks, development projects, etc.). Such short-term activities are fundamentally different from structural modifications of the river basin directly or indirectly linked to more permanent types of human activity such as deforestation, reafforestation, fire, land use changes, agriculture, urbanisation, etc., which often lead to long-term variations in sedimentary flux, and whose cause or causes can be both understood and evaluated [20,44].

Desperate attempts to identify relationships between concentrations, SM flux, and liquid flow rates seem hardly worthwhile when such impacts become primordial with respect to the sedimentary transit of a watercourse. From January 2005 to July 2006, the Isere was thus subjected to major development work on its flood-banks about one kilometre upriver from where this study's samples were taken, on the territory of the town of Gières. This had a significant effect on measurements, as the work resulted in considerable discharges of sediments over about 2300 m, with the building of a temporary track at the foot of the flood-bank. The track was made from gravel and silt scooped directly from the riverbed. The highest mean daily concentration (C_md) since the sampling instruments were set up was thus recorded on 18 April 2005, with an average of over 15.5 g L⁻¹ for the sector. The same year, the observed sedimentary flux was for the most part a direct result of anthropogenic disturbance, giving an annual transit some 7.5 Mt higher than what would have normally been expected (see Table 2).

To a lesser degree, work on building the Maurienne motorway in the Arc River valley begun in 1994, involved the consolidation of embankments and the building of civil engineering structures (in particular the Aiton, Saint-André and Escalade viaducts), and interfered with the assessment of daily sedimentary flux. This was especially true in 1994 and throughout 1995, when the work was mainly taking place in the area downriver from the Arc Valley, not far from its confluence with the Isere. In 1995, the recorded sedimentary flux was a great deal over the ‘normal’, most likely more than 2.5 Mt in comparison with an annual transit undisturbed by human activity (see Table 2).

For the above reasons, and in order to show the Isere's monthly characteristics as well as its seasonal SM transport model [20], the years 1995 and 2005 have been eliminated from the monthly and yearly statistics (Table 1). For the period from 1996 to 2004, the means, deciles and medians of concentration, as well as the mean flux, calculated from the daily figures, give a good indication of the Isere's sedimentary transport model (Table 1).

Seasonal variations in sedimentary flux are in relative agreement with the rain–snow regime of the Isere. If the concentration model shows up in the means (monthly variation index: 4.2), it becomes even clearer in the upper deciles, where the monthly differences are greater (monthly variation index: 5.9). In contrast, the mean specific flux, with a peak in May and a lesser one in March, is less characteristic of seasonal variations, as it depends too much on the sedimentary transfers linked to rising water levels. The maximum specific solid flow rate, for example, directly linked to rising-water periods, accounts for 47% of the variation in mean monthly solid flow, whereas deciles and medians are independent of such extremes. It is clear that days with a flux of around 63 t km² d⁻¹ (in one single day, 14 January 2004, a flux equivalent to 23,000 t km² yr⁻¹ was recorded) will certainly have an influence on monthly means.

Median and (especially) upper deciles, less influenced by extreme events, themselves independent of the seasons (three major rises in the river level were observed in May, March and January, respectively), seem the most pertinent factors to use in describing seasonal fluctuations of SM flux. They highlight the strong seasonal variations in sedimentary transit, superior to 15 (from 0.3 to 5.1 t km² d⁻¹) for the upper deciles of specific flux (Table 1). The months of May and June clearly show the importance of snowmelt in sedimentary production.

As with most rivers [21,29,36,37,42,43], the times at which the Isere rises are those accounting for most of its annual sedimentary transit. The evolution and quantities of suspended matter carried by the river during periods of rising water were closely monitored by means of twice-daily sampling. Since 1996, there have been three such periods at Grenoble, with an expected occurrence of 10 years or slightly more: in May 1999 (Q_max = 809 m³ s⁻¹), March 2001 (Q_max = 875 m³ s⁻¹), and January 2004 (Q_max = 750 m³ s⁻¹). With transits of over 1 million tonnes in ten days for the first two and half a million tonnes in four days for the third, these three episodes were typical examples of the importance of this type of event. They accounted for 34%, 39%, and 30% respectively of the annual tonnage (equal to 3.2 Mt in 1999, 3.0 Mt in 2001, and 1.7 Mt in 2004) for 1.3%, 5.4%, and 3.3% of the annual water volumes carried by the river. Rising-water periods thus require fairly regular monitoring of SM and show that estimations of sediment transfer based on insufficient or insufficiently frequent sampling may be somewhat unreliable.

Duration curves [21] show how important rising-water periods are, and enable the Isere's sedimentary variations to be more accurately discerned (Fig. 2). Apart from 1995 and 2005, the curves are all fairly similar, and can be superimposed on each other, with all of them remaining within the delineations of 2000 and 2004 (Fig. 2). Over half the Isere's annual sedimentary transit (from 44% to 71%, depending on the year; 55% on average) is generally transported within a lapse of 30 days, i.e. about 8% of the year. These figures partially agree with, and complement, Meybeck's estimations for the Rhone–Alpine area [20]. The curves also show that 1995 and 2005 were anomalous years, and thus provide an additional means of detecting the possibility of transit having been significantly influenced by work along riverbanks or in their immediate vicinity.

In order to make up for incomplete data (September–December 1998) and to be able to estimate sedimentary transits over a longer period of time, a relationship between SM flux and the flow rate of water needs to be established. Relationships such as SM = a Q^b, although regularly used in many studies [3,5,12,20,29], are poorly demonstrated (with a few exceptions) by daily data from the Isere, even if attempts are made to do so on a monthly or seasonal basis. The often-clockwise hysteresis of rising-water events is the main explanation for this, as sediments deposited during the previous such event go back into suspension when water level next rises [20,37,42]. Setting aside rising-water events, the use of regular, frequent sampling can lead to a large number of different cases, various types of high-water hysteresis, and differing pre-rise basin conditions. Irregular relationships between daily concentrations and observed flow rates are therefore hardly surprising when they concern a relatively large drainage basin located, moreover, in a high-mountain area, where processes of sedimentary supply, contributory zones, and pre-flood basin conditions may vary considerably over time and are even partly independent of seasonal influence.

In contrast, taken month by month, mean concentrations and, to an even greater degree, solid monthly flux (Q_sm in tonnes month⁻¹) are quite well related to monthly liquid flow (Q_m, in m³ s⁻¹); this is also usually given in the form of a power equation Qs=aQmb [3,29,30]. The relationship used for the Isere reads (Eq. (2) and Fig. 3):

The year 1995, despite improving the explanation of the model very slightly (r = 0.82), has been excluded. On the other hand, Fig. 3 clearly shows and confirms the influence of the work on the Isere embankments in 2005, with monthly figures for sedimentary transit that are clearly different from other years’ ones. The relationship has thus been established from the 1996–2004 period, and it explains around 66% of the variation in the Isere's monthly sedimentary flux. Months when water levels rise significantly, resulting in heavy transport, are nevertheless still underestimated. Such underestimation of transit is often unavoidable, even with strictly daily models, when high-water events have not been measured [3,4,29]. According to Eq. (1), underestimation of monthly figures is around 1 Mt for the rising-water events of May 1999 and March 2001, and around 0.5 Mt for that of January 2004.

Since the monthly figures described above take little account of periods during which the river rose to very high levels, it seems worthwhile to attempt to complete the present estimations of monthly transit by including days of very high flow rates. During flood periods from 1996 to 2004, power relationships, with daily flux and flow rates, match better than for the rest of the time, especially for flow rates in excess of 500 m³ s⁻¹ (r² = 0.65, with figures for 16 days, Eq. (2)), and steadily decline if flow-rate thresholds over 450 m³ s⁻¹ (r² = 0.52 with records for 36 days) or 400 m³ s⁻¹ (r² = 0.49, with 60 days) are used instead. For days when the daily flow rate is over 500 m³ s⁻¹, the daily transit can be modelled by Eq. (3):

Strictly speaking, the basic monthly transit (Q_sm, Eq. (2)), including daily flow rates over 500 m³ s⁻¹ only, should be estimated based on a new monthly flow rate. This approach has not been adopted, however, as its effect on final results is minimal, and it unnecessarily and probably deceptively complicates estimations of transit. For the three years in which a major rising-water event was observed, taking the sedimentary transit of episodes over 500 m³ s⁻¹ into account raised estimations of annual transit from 2.2 Mt to 3.1 Mt in 1999 (3.2 Mt were estimated from measurements), from 2.3 Mt to 2.9 Mt in 2001 (3.0 Mt measured) and from 0.8 Mt to 1.0 Mt in 2004 (1.7 Mt measured).

Overall, for the period between 1996 and 2004, taking into account, if only partially, such rising-water events over the threshold of 500 m³ s⁻¹ thus clearly improves the accuracy of final estimations of monthly sedimentary transit. The relationship between calculation and measurement of flux is improved, with a Bravais–Pearson coefficient of 0.82 instead of 0.62, when an estimation does not include daily flows of 500 m³ s⁻¹. The period for which data are missing, from September to December 1998, can thus be reconstituted in monthly stages with a certain degree of confidence in as much as the Isere was generally fairly low during that time. For the years measured, apart from 1995 and 2005, annual SM flux ranged from 0.89 to 3.16 Mt yr⁻¹ (Table 2), with monthly fluxes between 0.012 Mt (January 2002) and 1.78 Mt month⁻¹ (in May 1999).

As reconstituted since 1960, fluctuations of monthly and annual flux have increased slightly in amplitude, while remaining within the same general scale of values. Thus, over the 46-year period studied, the monthly flux varied by a ratio of 1:333, compared with a ratio of about 1:13 for the annual flux. These ratios naturally reflect the importance of rising-water events for sedimentary transit, but above all indicate considerable seasonal variations, intra-annual fluctuations being considerably greater than interannual ones.

For the period going from 1996 to 2004, the mean measured SM flux was a little over 1.54 Mt yr⁻¹, i.e. a specific flux of 270 t km^–2 yr⁻¹. The period studied nevertheless shows a mean model for the Isere slightly lower (167 m³ s⁻¹) than the one calculated over 46 years (183 m³ s⁻¹). Over the period from 1960 to 2005 (Fig. 4), mean SM export by the Isere, according to the estimations derived from reconstitutions (Q_sm + Q_sd), amounted to 1.76 Mt yr⁻¹, i.e. a specific flux of 308 t km⁻² yr⁻¹. Based on a larger number of samples and figures for sedimentary flux given in monthly steps over 46 years, this mean figure thus gives a more accurate revised estimation than that of 2.20 Mt yr⁻¹, reached in a previous study [10].

When solid flow rates of the Isere and the mouth of the Rhone are compared, based on a reconstitution of the Rhone's ones using the model by Pont et al. [29], both rivers show fairly similar annual transit variations (Fig. 4). From 1960 to 2005, the Isere contributed an average of around 11% of the volume of the Rhone's annual flow as measured at Beaucaire, but an average of nearly a quarter of the Rhone's annual sedimentary transit. However, the latter figure hides considerable interannual variations, on the one hand, and a clear tendency towards a drop in the Isere's contribution to the Rhone's SM sedimentary transit, with an average of 35% in the early 1960s, compared with an average of 22% at the beginning of the 21st century, on the other hand.