NONPOINT POLLUTION AND MANAGEMENT OF AGRICULTURAL AREAS - PHOSPHORUS AND NITROGEN TRANSFER IN AN AGRICULTURAL WATERSHED

Reduction of non-point pollution from agricultural land implies the knowledge of fluxes of pollutants, their origins and transfer mechanisms (CCE, 1974; Pluarg, 1978; CIPEL, 1985). To document the different processes involved, water quality of a small experimental watershed has been studied for 3 years. This basin (14 ha) is located in the catchment area of Lac Leman (Fig. 1, Table 1). It stretches across glaciary deposits in an agricultural area with grasslands and cereals. It represents a sub-unit of a larger rural watershed (33 km(2)) which has also been monitored in a previous work (Pilleboue and Dorioz, 1986). The water from the experimental basin flows into 3 ha of wetland. The mean water discharge at the outlet is 61 s(-1), with a maximum of 1001 s(-1) (Fig. 2,Table 2). The monitoring equipment used included: a rain gage, a limnigraph, automatic samplers, sediment traps (Fig. 1). Twenty storm flows have been monitored during the 3 years bf the survey. Samples were taken every 30 min and analyses were performed on 2 h cumulative samples. During low flow periods, a water sample was collected every week. Nutrient export coefficients measured (Table 3) were 14.6 kg ha(-1) year(-1) for N (22% of inputs) and 0.6 kg ha(-1) year(-1) for P (1.7% of inputs). One half of the total P was transferred as dissolved P and 73% of the total N, as nitrate. These annual mean values are relatively low compared to data previously found for large and intensively cultivated areas (CCE, 1974; Olness et al., 1975). The fluxes from the watershed were strongly variable (from 0.1 to 36 kg week(-1) for N and 0.005 to 0.3 kg week(-1) for P). P is transferred essentially during storm flows, while the N losses depended mostly upon the season (Fig. 2). The mean values characterizing the water of the experimental watershed are always below the agreed standard level for N but almost equal dr greater for dissolved P (Table 4). However the values are strongly related to the type of hydrological event (Table 6). Comparative study of the storm flows provided a good relationship between hydrological processes and variations in water quality: (1) Storm hows induced initially by a ''piston flow effect'' on the water table and followed by a limited run off. In this case losses were relatively small both for N (Fig. 4) and P (Fig. 3). Suspended matter and total phosphorus peaks coincide with the water peak and no lag effect was observed. (2) Storm flows with a significant runoff(Fig. 5). The peaks of suspended matter and phosphorus occurred at the very beginning of the storm flow and a lag effect appeared. Then, the total phosphorus concentration reached a very high level (up to 1 mg l(-1)) whereas nitrates remained at a very low level with sometimes a dilution effect during the water peak. (3) Storm flows with an important subsurface runoff (Fig. 6). P and N were essentially transferred in dissolved form. Extraction of N from soil was maximum and P was released in soluble form as several successive peaks. This phosphorus was transferred with the subsurface water flow (Fig. 7).