| | | 3.3 Monitoring of nutrient losses -- methodological aspects |
Within the context of the water protection work (e.g. the EU Water Framework Directive) it is necessary to know the sources contributing to reduced water quality in addition to their relative contribution within the watersheds. In the selection and implementation of optimal and cost-efficient measures, a proper understanding of the underlying processes resulting in e.g. nutrient losses is paramount. This understanding should form the basis for policy decisions in relation to water protection and land use issues.
Quantification of diffuse nutrient losses can either be done by direct measurements (monitoring) or by modelling. However, modelling can only be considered a tool which can supplement, and not replace monitoring. Measuring nutrient losses at catchment scale is not necessarily a straightforward task. The choice of methodology regarding the accuracy and precision of the collected data has to be a compromise between costs (e.g. equipment and operation and maintenance costs) and end-users demand (e.g. researchers, managers, public authorities). There is no specific method which can be recommended, except that each case has to be considered separately on the basis of the site specific conditions like catchment size, topography, channel characteristics, climatological conditions and the geo-hydrological settings.
Correct measurements of N and P losses (loads in surface waters at catchment scale) require reliable and precise data on nutrient concentrations as well as on water discharge. The reliability of the data may be affected by both field procedures as well as analytical procedures. The difficulties in obtaining correct water discharge data may often be underestimated. Different methods can be used to obtain information about stream discharge but they are often based on the combination of direct measurement of the water level and a known head-discharge relation for the measurement location, representing the relation between the water level and the discharge [122]. The measurement location can be a natural profile or fixed structure. In case natural profiles are used for discharge measurement, a head-discharge function has to be established though calibration measurements. However, a problem often associated with natural profiles is the changing flow conditions due to variations in roughness conditions in the channel bed because of vegetative growth. This can lead to large uncertainties in the head-discharge relation. In case a long term monitoring programme is to be initiated, it is an advantage to use fixed devices for discharge measurement. The main advantage of a such structure is the availability of a known head-discharge relation while in addition much higher accuracies can be obtained when they are build according to the specifications. It goes beyond doubt that a good discharge measurement program only can be obtained when sufficient attention is paid to the operation and maintenance of the measurement location and systems involved in the data collection and processing.
The selection of the appropriate water sampling strategy is an important issue in catchment monitoring programs [122]. In many cases the sampling strategy has to be adjusted to the dynamics involved in the loss processes thereby to avoid erroneous results. The losses of P, and to a lesser extend N, are typically event based, and depend on a combination of the prevailing geo-hydrological settings, climatological conditions and agricultural practices. The losses may vary greatly over relatively short time periods. In general one can differentiate between two sampling strategies being 1) grab or `point' sampling and 2) volume proportional composite sampling. A combination of these two sampling techniques can occur, for instance point sampling during stable, low flow periods and volume proportional composite sampling during certain storm events.
Comparative studies of different sampling strategies in the Nordic countries revealed substantial differences in calculated loads depending on the method used (e.g. Rekolainen et al., [388], Kronvang and Bruhn, [273]; Deelstra, [121]; Deelstra, [120]; Haraldsen, 2002). The general conclusion is that point sampling strategies tend to lead to erroneous estimates, particularly for the P losses. A Norwegian study [145] compared weekly point sampling during one year with flow proportional sampling in five small catchments. Weekly or forth-nightly sampling resulted in severely underestimated loads (<10% of the `true' figures for P as defined by the flow proportional sampling) as well as severe overestimates (more than twice the `true' figure). The deviations of N were generally less than for P. One should, however, bear in mind the prevailing geo-hydrological conditions in the Norwegian catchments, with a low base flow component in the total runoff volume in addition to many event flow situations of rather short duration.
| | | 3.3 Monitoring of nutrient losses -- methodological aspects |