Tier 2: Assessment Using the Riverine Load Approach

The riverine load approach (RLA) as presented in EC (2012) is based on data measured on site, that is, for the water, the suspended solids, river discharge as well as monitoring data from relevant point sources, and it calculates the basic processes of transport, storage or temporary storage and degradation of substances. The resulting riverine load provides quantitative information about the recent status of loading and, provided long-term information is available, also about trends over time (see Figure 7.3). In particular, it allows the allocation of observed river loads to point and diffuse sources (i.e., a basic source apportionment). If a reservoir or a lake is fed by different rivers, RLA needs to be implemented for each one significantly contributing to the nutrient inputs into the reservoir or lake. High nutrient concentrations, an increasing trend, or a high relevance of diffuse sources indicate a need for a more detailed analysis using the approaches in tiers 3 and 4.

The nutrient load transported by a river is estimated by taking the product of the mean flow-weighted concentration and the total river flow, expressed by the following formula (OSPAR, 2004a):

Figure 7.3 Utilisation of riverine data: Trends of total phosphorus concentration (flow adjusted in pg TP/Lxyear) in surface waters in Upper Austria for the periods: (a) 1990-2000 and (b) 2001-2004. (From Zessner et al„ 2016.)

L_Y=annual load (t/yr)

Q_d=arithmetic mean of daily flow (mVs)

Q_Meas=arithmetic mean of all daily flow data with concentration measurement (mVs)

C,=concentration (mg/L)

Q,=measurement of daily flow (m Vs)

Uf= correction factor for the different locations of flow and water quality monitoring station

w=number of data with measured concentrations within the investigation period.

Periods of high river flow typically carry a disproportionately large amount of the annual load of a contaminant (Zessner et al., 2005). To avoid underestimation of annual loads, it is therefore important that water quality sampling strategies are designed to capture periods of high river flow (Zoboli et al., 2015). Sites selected for sampling should be in a region of unidirectional flow in an area where the water is well mixed and of uniform quality. Both the particulate and soluble load of a contaminant should be quantified.

Flow Normalisation to Avoid Misinterpretation of Causalities

Riverine nutrient loads and, in particular, certain diffuse source components vary strongly with rainfall and hence river flow; typically, the wetter the year, the higher the load. Without the application of flow normalisation procedures, natural interannual variations in flow can mask or lead to misinterpretation of trends in nutrient loads. Genuine reductions in nutrient inputs attributable to the implementation of measures, for example, can be masked by the occurrence of higher annual river flow during more recent monitoring. Conversely, an apparently declining trend can be incorrectly attributed to the success of measures, but in reality reflects a drier year or years. Flow normalisation addresses this issue and can be undertaken via a variety of methods. Harmonised flow normalisation procedures are given by OSPAR (2004a). An example of a trend analysis of P concentrations in a river under consideration of flow normalisation is given by Zessner et al. (2016).

Estimation of Diffuse Loads

As discussed above, riverine loads can be used to calculate diffuse and unknown inputs of nutrients providing point source information is available. In the most basic approach, the diffuse load can be estimated as the difference between the total load (measured from river discharge multiplied by concentrations; see above) and the load discharged from point sources, as follows:

where, for a given contaminant, L_Diff is the anthropogenic diffuse load, L_yr is the total annual riverine load, and D_P is the total point source discharge. Such an approach ignores any potential in-river processes such as sedimentation and remobilisation, but provides a useful approximate estimate of the diffuse load of a given substance.

A more detailed formulation will be necessary where processes in the river or stream and natural background loads are thought to be significant. The following formula is based on an approach established by OSPAR (2004b) for the calculation of diffuse nutrient loads; in-river nutrient processing is typically significant:

where, for a given contaminant, L_B is the natural background load of the contaminant, and N_P is the net outcome of in-river processes upstream of the monitoring point. There are several methods to estimate N_P on a catchment scale. For example, Vollenweider and Kerekes (1982) derived a formula which described the relationship between the nutrient concentration at the inflow of a lake or reservoir and the concentration within it based on the water residence time. This formula can be used to calculate the retention of nutrients by in-lake processes (N_P for lakes). Behrendt and Opitz (1999) proposed something similar for rivers at a catchment-scale level. They derived a relationship between area-specific run-off (river flow subdivided by the area of the catchment) and nutrient retention induced by processes within the stream or river as well as a relationship between hydraulic load of a river (river flow subdivided by the surface of the water- bodies in the catchment) and retention. If the flow of a river, the nutrient load, the catchment area and/or the surface of waterbodies in the catchment are known, this approach can be used to estimate N_P for rivers on a catchment scale (OSPAR, 2004c).

The riverine load approach (RLA) provides a useful means of estimating diffuse inputs and/or validating modelled predictions. However, diffuse inputs from different sources are merged into a single value and are not, for example, distinguished between inputs arising from agriculture and those arising from urban run-off.