Approche expérimentale par crue artificielle de la participation du réseau hydrographique àl’exportation de phosphore par un petit bassin versant rural

Most of the phosphorus moved from watersheds to surface waterbodies is particulate and its transfer is in-stream processes dependant (Newbold et al., 1983). Phosphorus retention in the hydrological network can be due to settlement, biological consumption, chemical precipitation (bibr id=Pilleboue and Dorioz, 1986), adsorption (Green et al., 1978) and storage in the interstitial water (Garban et al., 1995). This retention is proportional to phosphorus concentration and is also dependent on hydrology (Dorioz et al., 1989; Svendsen et al., 1995). Inversely, most of the phosphorus is emitted during floods when soil erosion, resuspension of non consolidated stream bed sediments and stream bed erosion occur. The general objective of the work is to study the role of the hydrological network on phosphorus transfer in a small rural watershed during a minor flood. A mass balance between phosphorus input and output in the river (Dorioz et al., 1989) is difficult to establish during rainy periods because of the great number of phosphorus sources and their diffuse emissions. That is why an artificial flood (Casey and Farr, 1982) was experimented, using the addition of water at one location in a small river. The chosen river is located on a small rural watershed (310 ha) in the east of France, near Lake Geneva ([Fig. 1] and [Fig. 2]). The experimental reach is 300 m long and mostly covered with fine, non consolidated sediments (Fig. 3). There is no point source pollution entering the stream. The dryness of the experimental period allowed an easy increase in the flow from 5 to 20 l/s. The peak flow reached corresponds to summer high flows. Due to a high SO4 concentration, the injected water could be traced (Table 1). The strength of injected water (a strong jet of drinking water expelled out of a pipe over 2 h) made turbidity increase at the input point. In the reach (Fig. 3), water was sampled at the extremities (A and D) and at two intermediate places (B and C). Suspended sediments were measured at the four locations so that zones of resuspension and settlement could be detected. Total phosphorus, soluble phosphorus (<0.7 μm), conductivity and anions (Cl, SO4) were measured at the entry and the exit of the reach. The main results presented in this paper are: In spite of the increase of flow during the flood, the reach could be considered as a sink for particles and phosphorus (Table 2) entering at point A, even if the third part of the reach (from C to D) acted as a source (Fig. 5). The different shapes of the stream bed are responsible for such an opposing pattern. The hydrochemical measurements at the outlet (point D) show that during the first 120 min following the artificial injection, the water quality has not changed although the flow has increased. Only after that delay, the increase in SO4 concentration indicates the arrival of the injected water. The injected water left the reach 300 min after the artificial injection was stopped. According to Verhoff et al. (1979), this way of water renewal is due to the kinematic wave induced by the increase of discharge. This succession of water led to the existence of two peaks of suspended matter (Fig. 6). The first was due to the increase of flow resulting in resuspension in the third part of the reach. The second, smaller, is associated with the injected water. Phosphorus emission follows the same pattern (Fig. 6). An early and short peak of phosphorus (mainly particulate) is followed, an hour later, by a second increase, of mostly soluble phosphorus. A detailed mass balance of phosphorus, comprising settlement, resuspension and direct transfer is summarized in Table 3Fig. 7. It shows that most of the phosphorus exiting the experimental reach (70% for particulate phosphorus) comes from resuspension. However, the levels of phosphorus released during the experimental flood were low compared to what could be expected in polluted streams, or streams containing an important biological component. A mass balance of soluble phosphorus between A and D revealed that half of its mass was retained in the stream. During low flow periods, such a retention is not very significant. This retention during a small flood may be explained by a sorption reaction on particles. The same patterns described during the artificial flood occurred during a real summer high flow. Two peaks of suspended matter and phosphorus (Fig. 7) were induced by (1) the increase of flow and (2) the arrival of runoff water (Fig. 8). The proportion of resuspended matter could be estimated (Table 4). This study shows that the establishment of a detailed phosphorus mass balance in streams requires frequent sampling at the beginning of storm flow events. It also shows that the most bioavailable phosphorus form (soluble phosphorus) can be stored even during a small storm flow. This leads to a net gain towards the surface waterbodies, since P quality is modified as long as the storage goes on, and the bioavailability decreased. Finally, we show that within-stream disturbance is an important contribution of the total phosphorus emitted during small floods.