Introduction

P. Webster, BSc, MSc, PhD, DIC (Member)*, J. R. West, BSc, PhD, CEng, MICE (Member)**, A. M. Gurnell, BSc, PhD, DSc***, G. E. Petts, BSc, PhD***, J. P. Sadler, BSc, PhD**** and C. F. Forster, BSc, PhD, DEng, DSc, CEng, FlChemE (Fe//owy****

The history of the Tame catchment closely mirrors the development of the West Midlands conurbation which it drains (Fig. 1). The river was once renowned for its fishing but, following rapid urban and industrial development, served as an open drain for most of the twentieth century. Poor water quality and an unsightly appearance have resulted in a catchment strategy of river concealment, rather than promoting access to the river corridor. This is exemplifiedby the rectangular concrete channels which constitute the main channel as the Tame flows for 3 km, parallel to the elevated section of the M6 Motorway through the north of Birmingham. However, progressive river-water quality improvements have provided an opportunity for rehabilitation of the river and its corridor. This opportunity has been recognised by the Catchment Management Plan(') and the West Midlands Local Environment Agency Plan (LEAP)(*).The area has also been the focus of support under the NERCt thematic programme URGENT (Urban ReGeneration of the ENvironmenT). There are two principal objectives of the project:

Abstract Experience and knowledge of river-restoration schemes are generally available at the reach scale. However, there are problems with the application of this knowledge at the scale of large urbanised catchments which relate to understanding the system and predicting the impact of management strategies. This paper considers problems of developing a perceptive and efficient rehabilitation strategy for a large urbanised catchment, with specific reference to the River Tame in the West Midlands. Consideration is given to (a) rainfall-runoff relationships, (b) assessment of water quality and ecological status, and (c) predicting the impact of various management strategies.

(i) The scientific objective is concerned with identifying controls over the river's ecology. Specifically, the project seeks to investigate the relative importance of the boundary conditions of flow and water-quality regime, together with habitat, on the ecology of urban rivers. The various interactions, which are implied by this hypothesis, are illustrated conceptually in Fig. 2; and (ii) The project seeks to identify particular actions to improve the River Tame ecology. The study area is limited to the catchment draining to the Lea Marston purification lakes which facilitate on-line purification. It is referred to as the Tame catchment in the remainder of the paper, even though the catchment extends beyond this point.

River Flow Regime

Key words: Catchment response; river rehabilitation; River Tame;

urban rivers.

This paper was presentedat the CIWEM Rivers and Coastal Group Winter Meeting, held on 28 January 2000 at London Zoo, UK. 'Consultancy Services Manager, Hydro-Logic Ltd, Bromyard, UK. "Reader in Coastal and Water Engineering, School of Civil Engineering, University of Birmingham, Birmingham, UK. "'Professor of Physical Geography, School of Geography and Environmental Sciences, University of Birmingham, Birmingham, UK. '""'Lecturer, School of Geography and Environmental Sciences, University of Birmingham, Birmingham, UK. ""'Reader in Public Health Engineering, School of Civil Engineering, University of Birmingham, Birmingham, UK.

0J.CIWEM I 2001 1 15

I

JulyiAugust

The hydrological impacts of urbanisationare well documented(3) and refer to the increased proportion of runoff caused by conversion of natural surfaces to relatively impermeable ones and the reduced response time of drainage basins induced by higher flow velocities of both piped and channel flow. These factors, in combination, normally result in increased flooding. Furthermore, unless they are supported by wastewater discharges, baseflows are reduced. The combined effect of these impacts on the high and low-flow regime steepens flow-duration curves, i.e. renders urban catchments more 'flashy'. River flow is monitored at seven sites on the River Tame (Table l ) , the sites on each of the tributaries being located close to the tributary's confluence with the Tame. The data confirm that the Tame catchment is unique in a UK context, in terms of the extent of urbanisation within the catchment. In comparison, the largest gauged Thames tributary catchment in the London area is the Brent at Monk's Park, with an area of 116 km* and URBEXT,ggovalue of 40%. The data from the Tame provide a valuable opportunity to test generally held views of catchment response. An estimation of the percentage of rainfall, classed as direct runoff, remains a critical issue in engineering hydrology, and a number of flood events have been selected to investigate their characteristics. To-date, percentage runoff and other event characteristics have been obtained for twenty events for the tNatural Environment Research Council

167

P. Webster, J. R . West, A. M. Gurnell, G. E. Petts, J. P. Sadler and C. F. Forster on

0 Scale lOkm L---.

Fig. 1. Upper Tame Catchment (EnvironmentAgency, 1998)

engineering Channel characteristics

characteristics

Fig. 2. Principal links between project components Table 1. River flow gauges in Upper Tame catchment Ref.

Flow gauge

Area (km2)

28081 28003 28004 28039 28066 28094 28108

Tame at Bescot Tame at Water Orton Tame at Lea Marston Rea at Calthorpe Park Cole at Coleshill Blythe at Castle Farm Bourne at Shustoke

174.2 405.7 801.8 74.1 119.7 199.4 46.2

SPRHOST (%) 42 40 27 33 31 5.2 27

34.0 34.2 36.3 34.9 39.6 39.1 38.2

N.B. Catchment descriptors obtained from the Flood Estimation Handbook(4)CD-ROM (Institute of Hydrology, 1999) URBEXllw refers to the urban area in 1990 based on the ITE Land Cover Map SPRHOST is the standard percentage runoff for a natural catchment

'68

I

0J.CIWEM 1 2001 I 15 I July/August

Development, Flood Risk and the Urban Environment: Experiences from the River Tame

@= a L

40^^

S " " 1I c I"

*

4

c)n

4 4

4 4 4.

+ 4

4

1 day Rainfall ,Return Penod = 2 years

?I/ 0

10

20

4

30

=34.5mm

40

-

50

60

Event rainfall (mm) _~-______

~

/served

-SPRHOST

Events

-----.FE~

Fig. 3. Event percentage runoff: River Tame at Bescot catchments which are listed in Table 1. The variation in percentage runoff (PR) with the event rainfall for Bescot is shown in Fig. 3. The relationship is characterised by a degree of scatter largely due to problems of estimating areal (area mean) rainfall. However, the PR shows a general increase with event rainfall to avalue of about 20 mm. For event rainfall in excess of this value, PR appears largely constant. Similar relationships have been identified for other catchments in the Tame basin. Fig. 3 also shows the design PR from use of equations in the Flood Estimation Handbookc4).The design value exceeds the event PR for most events, being closer to the PR for a natural catchment, shown by the SPRHOST value. The explanation for these low values of PR mainly relates to the configuration of the stormwater drainage system. A significant proportion of the Tame catchment drains to combined sewers which drain to the Black Country Trunk Sewer (BCTS) and terminates at Minworth sewage-treatment works, the effluent from which discharges to the Tame downstream from Water Orton. The capacity of the BCTS at the works is 25 m3/s; therefore it is possible for the sewerage system to exert a strong modulating influence over the response of the river. It is also postulated that the operation of combined-sewage overflows (CSOs) could give rise to the specific form of the relationship in Fig. 3. HYDROWORKS model runs have confirmed the varying partition of runoff as conceptualised in Fig. 4. Because the catchment contains about 500 CSOs, they are likely to exert an impact on the hydrological response of urban rivers. Unit hydrographs have also been derived for a small number of flood events from Calthorpe Park and Be~cot(~). From an analysis of rainfall and flow data, it was found that the Flood

Studies Report(@equation for unit hydrograph time-to-peak and assumptions of a triangular unit response were acceptable. Mohd S00m(~)also sought to investigate flood-frequency curves for four sites in Table 1. Flood-frequency curves for the catchments have a tendency to flatten for return periods in excess of five years, and this pattern has also been observed in the Tame and urban catchments in Liverpool(*). However, a major factor in the Tame catchment is likely to be the presence of numerous minor flow obstructions in the tributary channels (comprising small culverts and bridges) where the conveyance may be reduced by debris during flood events. These give rise to secondary flood-storage effects but, nonetheless, they are potentially important in attenuating the contribution to the main channel and therefore reducing peak flows. In overall catchment-management terms, there is a balance to be struck between tolerating relatively minor flooding problems within a catchment, to ease the risk of flooding on the main channel. From the analysis of the flood events on the River Tame, its response to flood events is different from that presented in textbooks. Urban areas can respond in different ways, depending upon their age, which impacts upon the stormwaterdrainage configuration and the presence of secondary storage effects. An analysis of large urbanised areas demands that some effort should be made to characterisethe urban area, if it departs from the contemporary style of separate drainage with limited secondary storage effects: future hydrological research will seek to develop simple models of catchment response to reflect these important features. The models (Table 2) will also be configured to reflect the impact of management strategies which are presented in Table 3.

- 50%

100% STORAGE! LOSS

10

SEWER OUTFLOW

-

4

/.

0

;' ,' '

20

30

40

Total rainfall per event (mm)

Fig. 4. RainfalYrunoff relationship: River Tame at Bescot 0 J.CIWEM

I

2001

I

15

I

July/August

169

50

60

P. Webster. J. R . West, A . M . G u r n e l l , G . E. Pelts, J . P. Sadler and C. F. Forster on

Table 2. Approach to modelling of individual components DETERMINISTIC

STOCHASTIC STEADY STATE Category Rainfall Sewers River flow River qual. River sed.

'eec=c=G=c=eeeeeee

Subject Observed storm Flow, qual., sed. Hydrology, routing DO,BOD etc Erosion, dep.

Model Data Data SSFM SSQM SSQM

~~

~

~

_

_

_

_

ee=e=e=e=eeeeeee= Subject Model Observed storm Data Flow, qual., sed. H'WORKS Hydrology, routing Unit hyd., lSlS DO, BOD etc MIKE11 lSlS Erosion, dep. Habitat Exp. tanks Velocity, weed ISIS, 2D c=tcetec=e'ec=ec=e Subject Model Observed series Data Flow, qual., sed. H'WORKS Rainfall-runoff CATCHMOD DO, BOD Mass bal., kinetic Regular survey Observations, PCA Regular survey Observations, PCA Velocity, weed, sed. PHABSIM, Rule based mod1

*******=3*3***

Subject Design storm

Model IDF curves

Design flood

Flood frequency

_

EVENT Category Rainfall Sewers River flow River qual. River sed. Inv.hrert. Habitat CONTINUOUS Category Rainfall Sewers River flow River qual. Vegetation Inv.hR Habitat

*=333=+***-3***

Subject Design storm Flow, Qual., sed. Design flood DO, BOD etc Erosion, dep.

Model FSR, Stormpac H'WORKS Unit hyd., lSlS MIKE11 lSlS

--*********a* Subject Model Time series (generated) Flow, qual., sed. SIMPOL, H'WORKS Time series (generated) DO, BOD etc SIMPOL, SIMCAT

PHABSIM

N.B. Event = Pertaining to storm/flood events, IDF = Intensity-duration-frequency curves, PCA = Principal component analysis, SSFM = Spreadsheet flow models, SSQM = Spreadsheet quality model

Table 3. Possible management strategies Type

Strategy

Reference

Sustainable urban drainage Baseflow contribution Dilution flows Flood storage Sewerage

Source Control (i.e. tanks)

LEAP (Issues 4 & 5 ) LEAP (Issue 14) LEAP (Issue 13)

Sewage treatment Pollution from urban areas Pollution from urban areas River corridors Habitat provision Channel modifications

Use Birmingham GW Flood attenuation and water quality improvement CSD performance (tanks and screens) STW modifications Surface runoff First flush Improved habitat Provide refuges Increase available habitat

River-Water Quality Regime The river-water quality monitoring network which is managed by the EA comprises (a) about seventy locations where water samples are routinely taken, and (b) a continuous water-quality monitor at Water Orton. Data are available at 15-min intervals for dissolved oxygen (DO), amm. N, pH, temperature, turbidity, and conductivity, and the monitor is alarmed to provide indications of low DO. Because of the diversity of activities within the catchment, it is not surprising that the water-quality problems are complex. The progressive consolidation of sewage-treatment facilities at Minworth has led to a reduction in baseflows throughout the Upper catchment. About one third of the dry-

170

,

1

LEAP (Issue6)

LEAP (Issue 6) LEAP (Issue 4) LEAP (Issue 4) LEAP (Issue 3) LEAP (Issues 1 & 4 )

weather flow (DWF) at Water Orton is derived from sewage effluent, with an unknown proportion from industrial discharges. However, the greatest threat is posed under wetweather conditions, and it is estimatedcg) that the daily wet-weather loads exceed the dry-weather loads for most catchments and determinands. This is consistent with the industrial nature of most of the catchment and the number of CSOs within it. The problems associated with wet-weather events were highlighted during a highly localised storm event over a portion of the catchment in 1995(1°) when 44 mrn of rainfall were recorded. The time-series for flow, DO and amm. N at Water Orton are shown in Fig. 5. The major impact of the event was the 0 J.CIWEM

I

2001

1

15

1

July/AuQust

DeveloDment, F l o o d Risk a n d the U r b a n E n v i r o n m e n t : E x p e r i e n c e s f r o m t h e River Tame

T 160

6 1

140

120 i100

-Arnm. N

=2 a

u)

80

0)

60

5

40

0

20

0 9-JuI-95 10-JuI-95 1 1-Jul-95 12-JuI-95 13-Jul-95 14-Jul-95 15-JuI-95 16-JuI-95 17-Jul-95

35: 3301

f 25 s m

-g

15 10

u.5

1

0 t-9411-95

~.

1C-Jul-95

11-JuI-95

12-JuI-95

13-Jul-95

14Jul-95

15Jul-95

1

-

I

16J~l-95 17-Jul-95

Date (hour)

Fig. 5. Flow, DO and amm. N: River Tame at Water Orton, July 1995 rapid depletion of DO in the River Tame, resulting in an estimated reduction in fish biomass of 97% between Lea Marston and the River Trent. The River Cole was also badly affected. The action of CSOs, providing a source of BOD, is often implicated in such events. However, an inspection of the amm. N concentrations for the event shows that CSOs (and by implicatlon, effluent) made only a marginal contribution to the first flood event on 10 July 1995. Subsequent field investigations by Environment Agency staff reported the influence of fine sediments (with high oxygen demand) in the bed of the River Cole. It has been observed that critical water-quality conditions are rarely associated with critical storm/flood conditions, i.e. quality problems cannot be tackled by some ‘design event’ approach. The application of continuous flow and water-quality models is clearly impractical in resolving such complexity. However, it would seem that the philosophy adopted in the SIMPOL procedures advocated in the Urban Pollution Management procedure(ll), has the potential to offer a pragmatic approach to the resolving of this complexity. As with the hydrological work, a key feature of the waterquality investigation is the development of simple models of catchment response, which can be used to investigate the impact of management strategies (Table 3).

Hydro-ecological Characteristics Whilst flow and water-quality regimes have important influences on river ecology, the physical characteristics of the river channel are also of fundamental importance in determining faunal communities. Tools have been developed for estimating the potential of less impacted rural rivers to support particular macro-invertebrate communities, e.g. RIVPACS, but these tools are not applicable to urban river channels. Therefore, this component of the project aims to identify the physical characteristics of urban rivers and to classify distinct river channel-types within the heavily modified channels of the Tame catchment which might support different ecological communities. The project work has involved: 0 J.CIWEM

I

2001

1

15

I

Juiy/August

(a) The development of a hierarchy of spatial elements of urban river channels; (b) Definition of the distinctive properties of channels at each scale within the hierarchy; (c) Design of a database in which information on these properties can be stored: and (d) Classifications of urban rivers based upon an analysis of the data. The key spatial level is the stretch, because this is a length of channel which is subject to a single type of engineering modification. Other spatial scales represent aggregates (catchment, sector) or subdivisions (unit, habitat, patch) of stretches. The information accumulated from secondary sources (catchment and sector levels), purpose-specific surveys (sector level), and a combination of new and archive surveys (unit, habitat, patch levels) has been entered into a relational database using Microsoft Access. The database can be queried to address sitespecific issues or to support multivariate data analyses which can generate classifications of urban rivers. So far, the analysis has focused on the stretch level and contained physical habitats, e.g. bars, pools, riffles. Five classes of urban river stretch have been identified (semi-natural, lightly modified, modified, moderately modified, heavily modified), reflecting boundary material, planform and bankbed reinforcement. Clear but unexpected patterns in the frequency and diversity of physical habitat types have been identified within the five classes of urban river stretch(12).These results:

(if Underline the fact that there is a substantial variety in the physical characteristics of urban rivers, and therefore potential for investigating the complexity of their hydroecology; (ii) Demonstrate a clear link between physical features at different spatial scales (stretch and habitat); and (iii) Suggest that, even in urban river channels which have been heavily modified by engineering intervention, a diversity of types of physical habitat can still be supported. 171

P. Webster, J. R . West, A. M. G u r n e l l , G. E. Petts, J. P. Sadler and C. F. Forster on

The study of macro-invertebrates within the Tame catchment aims to develop biological methods for assessing environmental quality for application in urban watercourses. The large number of potential contaminants from diffuse and point sources makes comprehensive chemical analysis either impossible or prohibitively expensive. Furthermore, the critical concentration (above which a chemical may cause significant ecological harm) is often unknown, while synergy between chemicals might have greater ecological impact than could be predicted. The sampling design aims to discriminate between the influences of water quality, flow regimes, and physical habitat. Therefore, macro-invertebrate sampling has focused on river channels within the Tame catchment which are subject to differing flow and water-quality regime. Within these channels, macro-invertebrate sampling has been undertaken within paired, adjacent river units and habitats which possess a range of different physical characteristics. The variety of species and environmental tolerances of Chironomids(13) makes them particularly useful as environmental indicators in urban environments. Field-survey work and subsequent laboratory work has helped to identify the distribution and deformities of Chironomid larvae at the sampling sites on a seasonal basis.

Modelling Strategies There are two principal objectives to the modelling work: (i) the modelling is an essential component of scientific enquiry, providing a basis for hypothesis testing, and (ii) it provides an opportunity for developing predictive relationships for managing urban catchments. The models need to reflect the broad components whose interactions are illustrated in Fig. 2. This represents a significant challenge, in view of the complex interactions, which vary in time and are influenced by a combination of local effects and upstream catchments effects. This complexity is revealed by Table 2, which shows the range of models being used to simulate the different components in Fig. 2. A key use of the modelling tools will be to investigate the impact of various management strategies on the ecology, and an outline of these actions is presented in Table 3. Two broad approaches, i.e. statistical and deterministic, evolve to meet this need to predict the effects of management impacts. Modelling Approach: Statistical The previously described database can easily be queried. In particular, data sets can be extracted for analysis at single or multiple-spatial scales. Complexities of the large number (varied nature and interdependency of the hydrological, hydraulic, water quality, physical habitat and biological variables) make simple bivariate analyses of the data inappropriate. Instead, a range of multivariate techniques is being applied. A combination of ordination (e.g. principal components analysis, detrended correspondence analysis) and classification (cluster analysis and TWINSPAN techniques) are being used to investigate structures and associations within the data. For example, the initial classification of urban river stretches was established using cluster analysis. Such analyses are capable of identifying key processes and site variables, indices and classes, which are associated with particular ecological characteristics, and thus may support management decisions. In addition, it might be possible to develop predictive equations which managers can use to explore the impact of making changes to any of the boundary conditions shown in Fig. 2. 172

Modelling Approach: Deterministic An alternative modelling approach, based upon a deterministic understanding, is far less direct and demands an appreciation of the physical, chemical and biological factors which impinge on the ecological potential of a site. The influence of flow regime and physical factors is already addressed by the ‘physical habitat simulation system’ (PHABSIM) which has been applied to numerous sites in the UK, but these have been mainly on non-urban sites. An important part of the study will be the modification of the existing methodology to provide a system for incorporating the effects of water-quality regime into PHABSIM, therefore making it more applicable in the urban environment. Overview of Modelling It is axiomatic that each modelling approach leads to advancement towards the scientific objectives of the project. Of interest to the wider community is the extent to which the models can be used to investigate the impact of management strategies. The statistical approach could provide a rapid means of appraisal for large catchments; however, the work completed to-date on the catchment has demonstrated the complex manner in which it responds to environmental and anthropogenic changes. The extent to which the statistical approach is able to reflect this complexity is not clear. Stringent validation procedures will be required for any models arising from the statistical approach to ensure that they have sufficient predictive capability. The rigour of the deterministic approach should cope better with the complexity of the physical, chemical and biological processes. However, the application of the component models to the urban environment usually requires extensive resources; in particular, the specification of relevant model boundary conditions represents a distinct challenge.

Conclusions 1. The hydrological response of historically urbanised catchments does not accord with generally held views of urbanised catchments. The storrnwater drainage system, and the secondary storage within the catchment, can exert a powerful modulating influence over the flood response. 2. Water-quality regimes are subject to even greater complexity in historically urbanised catchments. In particular, critical water-quality conditions are seen to occur for moderate storm events. An analysis of continuous timeseries is essential to advance the understanding of stochastic and deterministic controls over river-water quality. 3. The establishment of a comprehensive database describing the hydro-ecology of the river network and its corridor is an essential prerequisite to urban river management. The derivation of various purpose-specific classifications of urban rivers is an important building block in the understanding of urban catchments. 4. An investigation of catchment-management strategies demands the application of appropriate simulation models. To be of operational value, these need to incorporate the principal physical, chemical and ecological mechanisms which prevail in urban catchments. It remains a significant challenge to combine these in such a way as to achieve adequate fidelity of models, within the constraints of practicality and data availability. 5. There is pressure to meet this challenge, due to public demand for environmental improvement and the associated 0 J.CIWEM

I

2001

I

15

I

July/August

Development, Flood Risk and the Urban Environment: Experiences from the River Tame

limitation c financial resources. Accordingly, there is an unquestionable need for a reliable decision-support system that catchment managers can use to target funds efficiently to enhance the urban environment.

Acknowledgements The work described in this paper relates to a collaborative research project between the University of Birmingham and the Centre for Ecology and Hydrology. The project is funded under the NERC URGENT programme, with additional funding from the Environment Agency. The authors wish to acknowledge the contribution of members of the Project Steering Group, including representatives from the Environment Agency, Severn Trent Water, and local authorities.

References (1) NAnoNnL RIVERS AUTHORITY. Tame Catchment Management Plan: Consultation Report. 1996. (2) ENVIRONMENT AGENCY. Local Environment Agency Plan: West Midlands Tame. Consultation Report. 1998.

0 J.CIWEM

1

2001

I

15

I July/August

(3) SHAW,E. M. Hydrologyin Practice. Chapman and Hall. 99 . (4) INSTITUTE OF HYDROLOGY. FloodEstimation Handbook.Wallingford, UK. 1999. (5) MOHDSOOM,S. Flood Response of the Upper Tame catchment. Unpublished MSc Thesis. The University of Birmingham, 1998. (6) NATURALENVIRONMENTRESEARCH CouNcii. Flood Studies Report. NFRC lnsfifufe of Hydrology, Wallingford, UK. 1975. (7) JEREMYBENNASSOCIATES. Flood Hydrology of the Upper Tame. Report for EnvironmentAgency, 1999. (8) SHEMILT,S. J. The Catchment Hydrology of the River Alt, Merseyside. Unpublished MSc Thesis, The University of Birmingham, 1998. Initial Planning for River Tame UPM Investigations: (9) WATERRESEARCHCENTRE. Phase 1, Identification of Wet Weather Pollution Problems. Environment Agency (Midlands Region) 1998. (10) NATIONAL RIVERSAUTHORITY. Report on the Tame Incident: 10-12 July, 1995. NRA. 1995. (11) WATERRESEARCHCENTRE.Urban Pollution Management. Foundation for Water Research,1994. (12) DAVENPORT, A. J. GURNELL,A. M. AND ARMITAGE,P. D. Hydro-ecological classification of urban rivers. In froc. ofBritish HydrologicalSocietyAnnual Symposium,Newcastle upon Tyne, September 2000. ?I s.AND PINDER,L. v. C. (Eds.) The Chifonomidae. (13) ARMITAGE, P., CRANSTON, The BiologyandEcology of Non-Biting Midges. Chapman and Hall. 1995.

173