WFD48 Stage 3
March 2006


The WFD 48 project carried-out the work necessary to revise water resource regulatory standards covering abstraction and impoundments for rivers and lakes, throughout the UK based upon ecological status. This report details the results of Stage 3 of WFD 48, which determines the appropriate environmental standards (i.e. the required thresholds) for water resources parameters for UK river and lake water bodies.  The set of standards are appropriate to deliver the WFD and relate to the boundaries for all five WFD classification bands: high (HES), good (GES), moderate (MES), poor (PES) ecological status. The standards do not relate to other impacts on ecological status of rivers, such as physical obstructions to fish migration or temperature changes to riparian tree clearance.  Stage 1 of WFD 48 reviewed the appropriateness of different parameters, such as river flow and lake level, as environmental standards. Stage 2 reviewed the potential typologies of rivers and lakes for defining water bodies in which different standards would be appropriate.  Stage 3 produces the procedures for classifying any water body into the various types and defines the thresholds for the parameters within each type.

The UK has relatively few species and few people reliant directly on river resources for their livelihoods. Consequently, environmental standards for water resources can be quite simple. However, the resulting standards should not be used outside of the UK.

Developing a typology for river water bodies

The four elements of fish, macrophytes, macro-invertebrates and physical structure provide good indicators of river ecosystem status. The initial intention of the project was to adopt a typology for each element (e.g. RIVPACS for invertebrates). However, the only appropriate typology of UK rivers based on ecological data that can be used for water resources standards is that defined using macrophyte communities from 1500 sites (Holmes et al. 1998).  This classification was simplified to give 8 generic river water body types (A1, A2, B1, B2, C1, C2, D1, D2).  The types differentiate, for example, lowland, low gradient, clay substrate river water bodies (A1), lowland chalk streams (A2) and steep, upland, coarse-grained substrate river water bodies (D2).  Chalk streams were further sub-divided into headwaters and downstream areas. Expert consensus was that this typology was suitable for setting standards for macro-invertebrates and macrophytes.  The experts recommended the use of the 8 fish community types defined by Cowx et al. (2004) for setting standards for fish. In only one type (salmonid spawning and nursery areas), standards for fish exceeded those standards set for the 8 generic types (defined for macro-invertebrates and macrophytes). Consequently, only this fish type was used.

Physical river water body types

Since the environmental standards needed to be applicable to all river water bodies without any site visit, classification was based on variables that could be quantified using existing datasets.  To understand the variables that differentiate UK river water body types, catchment characteristics were derived for a set of 781 river flow gauging stations, giving good geographical coverage over the UK.  The catchment characteristics included topographical, climatological, soil, flow variables. Principal Components Analysis (PCA) was used to define uncorrelated linear combinations of these characteristics, which permitted the selection of a small number of dominant characteristics, that most strongly differentiated different river water body types. Rainfall, slope and altitude broadly differentiate north west UK (wet, steep, high) from south east UK (dry, low gradient, low altitude).  Drainage area differentiated water bodies with catchments of different scale. Base flow index (BFI) differentiate flashy from base flow dominated water bodies. Together these three components account for 61% of the variance in UK water bodies and can be interpreted intuitively. From the PCA, rainfall (SAAR), altitude (ALTBAR), slope (DPSBAR), drainage area (AREA) and baseflow index (BFI) were selected to characterise variation along the first three components and hence to provided a basis for discriminating between types.

Linking biology to physical river variability

The WFD System A catchment typology that uses altitude, catchment area and geology was examined, but the broad classes meant that it did not discriminate between the generic river water body types in the UK.

Recursive partitioning analysis (Rpart) was employed to classify the generic river water body types according to the physical characteristics selected from the PCA. Rpart constructs hierarchical binary classification trees, giving a series of splits based on cut-points in the explanatory variables.  SAAR, AREA and BFI were able to predict membership of all classes except C1 and D1. Class C1 contains relatively few sites which have a wide geographical distribution and are not individually distinctive sites. Consequently it was of crucial importance for the model to differentiate this type. Class D1 sites have specific locations in English Lowland Heaths (e.g. New Forest), Scottish Flow Country and Western Isles and can thus be located geographically.  Separate analysis was conducted for different hydro-eco-regions, but this did not improve predictive power in the model.  The use of substrate data was also explored. Some general patterns emerged justifying future research, but no significant improvement in predictive power was gained.  No method was available to define salmonid spawning and nursery areas, other than use of local knowledge.

Maps were produced to demonstrate how river water body generic types vary along major UK rivers from, for example, D2 in the headwaters to B2 near the mouth for the Rivers Tweed and Exe.

Defining environmental standards for river water bodies

Regulatory standards for each river water body type were defined through an expert consensus workshop approach. The experts were invited to define thresholds of flow alteration that would ensure good ecological status (GES) in water bodies, based on two scenarios: abstraction and impoundments. Experts in macrophytes and macro-invertebrates adopted the generic typology and defined thresholds for abstraction for each river water body type to achieve GES. The fish experts defined standards for fish community types. All standards were very precautionary based on indicating points at which experts could no longer be certain that GES would be achieved. The thresholds were broadly in the range of 10-20% permissible abstraction above flows of Q95 with hands-off below Q95; these stringent levels reflected uncertainty in precise threshold levels. All experts felt that standards for impoundments should be the same as those for abstraction and that long periods of constant compensation flow releases from impoundments could achieve Good Ecological Potential (GEP) but would not achieve GES, which requires maintenance of flow variability.

No precise method was available to identify fish community types for UK river water bodies. A fish atlas is available for Great Britain at 10 km grid scale, but this does not include spawning and nursery areas.  In general terms, fish community 1 (chalk river fish) relates to generic type A2 (chalk rivers); fish community 2 (eurytopic/limnophylic fish) relates to A1 (lowland clay-substrate rivers). The other fish communities  cut across generic types; rheophilic cyprinids could occur in types B1, B2, C1, C2 and adult salmonids and salmonid spawning could occur in types B1, B2, C1, C2, D1 and D2.  Future research is required to be able to predict these latter three fish community types in river water bodies. Recursive partitioning analysis (as used for the generic types) provides one possible approach.

Results of analysis of LIFE score data were not able confirm any variations in standards between river water body types so long as the flow regime was standardised by both mean and flow variation. Analysis of changes in physical character of river water bodies (based on wetted river width) reinforced the significance of Q95 as a threshold at which sensitivity to flow changes.

In all cases, except salmonid spawning and nursery areas, standards set for the generic types (based macro-invertebrate and macrophyte flow requirements) were more strict than those set for fish community types.  Consequently, only the salmonid spawning and nursery area type was retained as an explicit river water body type for WFD 48, in addition to the 8 generic types.

Practical standards, less stringent than those precautionary standards defined by experts (which are likely to “guarantee” a particular status will be met), were derived by the project team by taking a risk-based approach. This approach accepts that with more relaxed standards, some river water bodies may fail to achieve the desired ecological status, but these would be identified by appropriate monitoring. In this way the team defined standards for the lower limits to achieve different levels of ecological status:  

% of flow
 Lower limit for flow > Q95 Lower limit for flow < Q95
High Ecological Status  10  5
Good Ecological Status  15-35 7.5-20
Moderate Ecological Status   25-45 15-30

In general the strictest standards are those for steep upland rivers (D2) and chalk streams (A2) whilst the least stringent tend to be for lowland clay-substrate rivers (A1).   All standards were defined in terms of % of flow on the day abstraction. The experts felt that it was not possible to specify a constant volume that could be abstracted at any flow and still achieve GES, except by defining the volume as a percentage of the very lowest flow. As a fail-safe, it was suggested that any abstraction should not reduce flow at Qn99 by more than 25%.

Although not strictly part of the project specification, the team analysed the experts’ views on releases from impoundments that would achieve GEP. These included release of flood events at key times of the year and variations and fluctuations in compensations flows.

Defining a typology for lake water bodies

The fundamental approach to defining a typology for lake water bodies was to adopt the lake reporting typology adopted for Great Britain.  This is based essentially on chemistry, in turn reflecting geology and salinity, and giving basic classes for peat, low, medium and high alkalinity, marl and brackish waters.  It was recognised that the typology could allow for classes based on these types to be split or combined, in order to increase sensitivity to factors specifically relevant to the fundamental hydrological variable of water level, or to avoid duplication, respectively as appropriate.  Such an approach offers the advantage that supporting biological data are now being collected and analysed.

Physical lake types

Further to the chemical basis of the typology reflecting geology and salinity, a second tier of the typology reflects lake depth (two classes for mean depth less than or greater than 3 m), and further lower tiers are based on altitude (three classes divided at 200 m and 800 m) and lake size (three classes divided at water area 10 ha and 50 ha).  Strong control on the geographical distribution of these classes is exercised by geology: deep lakes are much more concentrated in the north while shallow lakes are much more abundant in the south.  A further physical characteristic is basin form, whereby a further two-fold division of classes has been proposed following the work of Håkanson.

Linking biology to physical lake variability

The typology has been developed on the basis that water level alteration is the principal hydrological parameter to which aquatic communities are sensitive, while the degree of sensitivity is a function of many other lake characteristics, as identified above.  The expert workshop identified that there were substantial deficiencies in the knowledge necessary to confidently predict the threshold hydrological alterations which would lead to changes in ecological status for lakes of various physical and chemical characters.  The incorporation of effects was achieved by using the chemical types to define basic levels of sensitivity, located by reference to the limited opinions expressed at the expert workshop and as refined according to the literature, and then defining sensitivity modifiers according to each of the additional factors outlined above.  Basic levels of sensitivity ranged from 10-20% deviation in naturally occurring lake levels.  A sensitivity calendar was used to identify and collate the seasonal nature of sensitivity effects for different species and groups of organisms.  A risk-based approach was developed, whereby the number of sensitivity-increasing factors applying to an individual lake in an individual season was used to identify the degree by which the basic sensitivity threshold should be reduced.  The principal threshold of interest was that representing the boundary between Good and Moderate Ecological Status, and was initially expressed as a proportion of naturally-occurring lake level on any day, relative to the sill or control structure over which the outflow drains.

Defining environmental standards for lake water bodies

To provide environmental standards in terms of water flows, allowing regulators to work towards licences in volumetric or flow terms, it was necessary to relate water level deviations to flows.  This was possible using the assumption that flow over a sill or other outflow is related to level by a rating relationship with a stage exponent greater than unity.  This is as indicated theoretically by the Chezy equation and has been confirmed empirically in this study by reference to data from a necessarily small number of sites at which levels and flows are available.  Assuming the Chezy exponent of 1.5, permitted abstraction fluxes were found to be more lenient than their corresponding (and more ecologically relevant) permitted water level deviations.  Level restrictions ranged from a mere 5% for some peat lakes to 20% for some brackish lakes, corresponding to abstraction restrictions of 7% to 28% respectively.

This system of defining environmental thresholds provided for individual differences attributable to specific physical controls to be reflected through the concept of risk, but led to results with an unjustifiable level of apparent accuracy.  As a means of addressing this concern, the final table of environmental standards, expressed in flow deviation terms, therefore introduces a rounding to the nearest 5%: the loss of unjustifiable minor differences in threshold is argued to compensate for possible exaggeration of threshold values in some cases where similar lakes fall just either side of a 5% boundary.  The set of thresholds for the Good/Moderate Ecological Status boundary was then taken as a starting point to define threshold values for the other ecological status classes, while maintaining the assessments of relative difference in sensitivity.

By ultimately defining standards in flow terms, it becomes possible to assess the possible effects of a water use proposal not only in relation to the adjacent rivers but also any lakes on the same river system, and so therefore be able to identify whether the river or the lake provides the more stringent environmental requirement.  Given the assumption of the Chezy equation applying to outflow ratings, it is likely that lakes will often require less stringent provisions than rivers.

Proposals for future work

Recommendations for future research are provided including: a method to predict which river fish community types occur in which water bodies; use of site variables such as channel geometry and substrate using RHS; applying environmental standards to licensing; involving researchers more closely with development of standards; and defining standards for flows to estuaries.  For the proper setting of environmental standards for lake abstractions, further recommendations are made, especially in relation to increasing the amount of monitoring of lake levels and their associated outflows.

Copies of this report are available from the Foundation, in electronic format on CDRom at £20.00 + VAT or hard copy at £35 .00, less 20% to FWR members.

N.B. The report is available for download from the SNIFFER Website