The influence of man on the hydrological regime with special reference to representative and experimental basins — L'influence de l'homme sur le régime hydrologique avec référence particulière aux études sur les bassins représentatifs et expérimentaux (Proceedings of the Helsinki Symposium, June 1980; Actes du Colloque d'Helsinki, juin 1980): IAHS-A1SH Publ. no. 130.
Urban runoff pollution in France: a national programme
MICHEL DESBORDES Laboratoire d'Hydrologie Mathématique, Montpellier, France JEAN-CLAUDE DEUTSCH Service Technique de l'Urbanisme, Paris, France
JEAN-CLAUDE HEMAIN Laboratoire d'Hydrologie Mathématique, Montpellier, France
Abstract. The Ministry of the Environment has recently set up a working group to study the pollution of urban runoff as a function of land use. The following three criteria have been deter- mined: experimental design, methods of sampling and measurement, and the choice of suitable urban basins. Four basins, two in the Paris region and two in the south of France, have been chosen. Each pair of basins is subjected to the same series of rainfall events. The measuring programme is due to begin in 1980.
Caractérisation de la pollution des eaux de ruissellement en France: un programme national Résumé. Un groupe de travail a été organisé par le Ministère de l'Environnement et du Cadre de Vie en France pour définir une méthode de travail permettant de caractériser la pollution des eaux de ruissellement urbaines en fonction des types d'occupation des sols. Trois éléments ont été déterminés: un plan d'expérience, un choix de bassins versants expérimentaux, une méthodo- logie des mesures. Quatre bassins versants ont été déterminés: deux en Région Parisienne, deux dans le sud de la France. Chaque couple de bassins est soumis à une chronologie d'événements pluvieux identique. La campagne de mesures doit débuter pendant le premier semestre 1980.
RESEARCH SUMMARY
The study of the pollution of urban runoff is quite recent in France. The first experimental study began in 1972, and, until 1977, a few others were undertaken for local purposes, as money was available. Since then, the Technical Service of Urban Planning and Management, of the French Environment Ministry, has taken care of the problem. This service is responsible for the coordination and technical assistance in urban planning and management for other services of the ministry and to municipal councils. As the results of the first studies became known, it seemed important to estimate which components of waste water and urban runoff had the largest polluting impact on receiving waters. This was an important question for municipalities; for example, the choice between a tertiary treatment of waste water and treatment of storm runoff.
As early results from other countries were relatively incoherent, and as the con- ditions of measurement were often not well-known, a working group was created with sub-groups to consider:
(1) characterization of urban basins and sewer layout, (2) methods of measurement and their interpretation, and (3) treatment of runoff pollution.
Only the first two sub-groups are working on the characterization of the pollution of runoff.
At the first working meetings the scope of the study was defined. The possibility of solving the problems using experimental data were confirmed, but the use of foreign data was excluded for the reason given above. A'literature review (Desbordes and Ribstein, 1978) has shown the validity of the latter point. It was also decided to use only direct measurements.
29
30 Michel Desbordes et al.
The measurements of runoff quality were organized in order to minimize as far as possible the effects of sewers (sedimentation and suspension in sewers, etc.) on some experimental urban basins. The method was well known in France as it has been used since 1974 in the estimation of urban rainfall—runoff relations. These measurements were used: (1) to find the total annual pollution in the receiving waters for a given land
use, and (2) to characterize the most important rainfall events for the pollution impact in receiving waters.
The group has concentrated on the definition of the following elements in order to obtain good runoff quality data: experimental design; methods of sampling and measurement; and a choice of good urban basins.
EXPERIMENTAL DESIGN
The aim of the experimental design is the a priori determination of the number and type of measurements needed to satisfy the two research objectives given above. It takes into account experience gained in other countries and relates only to runoff pollution. This limitation introduces strong experimental constraints. Moreover, the complexity of pollutant deposition and transport and financial constraints mean that some degree of hypothesis is necessary.
The constraints
These deal with the sewer system and the basin.
For the sewer:
(1) it must be separate, without false connections;
(2) hydraulic cleaning conditions must be satisfied;
(3) there must be no special works giving pollution upstream of the measuring point (sedimentation works, etc.);
(4) it must be possible to put samplers in the sewer at the measuring points;
(5) the measuring section must also be suitable for accurate flow measurement;
(6) the sewer network and the corresponding drainage areas must be well defined.
For the basin :
(1) it must have a central position (no edge effect);
(2) erosion on natural areas must be known and minimized;
(3) no street sweeping and cleaning;
(4) no significant building activity;
(5) no occasional activity giving significant pollution.
It would seem, therefore, that the basins must be quite large with non-homogeneous land use.
Description of the processes
For a given pollutant and a given basin, the deposition and transport of pollutant is demonstrated in Fig. 1. P rainfall events occur during the observation period T. For a
Ap ^ Ep
£2, tiA ^ TSP A p-
p rffi • lin r
•Tme0 , 1 tp
FIGURE 1. Deposition and transportation scheme.
Urban runoff pollution in France 31 given rainfall event p , the total transported mass of the Arth pollutant is Ep.Ap is the accumulated mass during the dry weather period between the p — 1 and p rainfall events. Rp is the residual pollutant over the basin after event p. Ra is the total mass over the basin at the beginning of the measurement period.
If the duration is long enough and if the conservation of mass of pollutant is satisfied then the R0 — (Ap + l +Rp) term of the general mass balance must be small (except in the case of a long dry spell) compared with the total accumulated mass MA(T) and the total transported mass ME(T). These two masses should then be
almost equal.
If the mass of the k pollutant is not conserved, Rp will not be completely trans- ported for the p + 1 events. SoME(T) will be less than MA(T) and the difference will depend on the chronology of the rainfaË events and Ap will after all be equal to the resulting pollution if that pollution stays over the catchment and if it is conserved.
The above scheme may be a basis for the modelling of runoff pollution. Ep is the only measured quantity ;AP andi?p are related to it by
Rp-i +Ap=Ep +Rp (1)
A complete description of the whole set is mathematically indeterminate. This is still more evident if the pollutant is not conserved for then equation (1) becomes
Rp-i (tp) + Ap(tp)=Ep +Rp(tp) (2)
Some experimental studies have shown that the direct estimate of Ap (by sweeping, for instance) is not very realistic. The determination of Ap will necessitate a number of samples to be taken over a small homogeneous area of the basin. If the pollutant is not conserved, samples must be taken to determine both Ap(tp) and Rp(tp)\ Moreover, the measurement of Ap would be of a differential type while it is an integral for Ep : the relationship between these two variables could only be found with much con- jecture concerning the production—deposition—transport process of runoff pollution.
The lengthening of the measuring period may perhaps be a solution, using a statis- tical analysis of Ep which may be explained for each basin by the rainfall and dry weather characteristics. The variation of some particular Ep values could then be studied in connection with land use and basin characteristics. Such an analysis would necessitate a large number of basins and so would be quite expensive.
Therefore in the first instance, a reduction of the number of measurements requires the statement of some hypotheses and to their confirmation over a short period. If they cannot be confirmed, a lengthening of the measuring period could perhaps lead to a solution of the problem.
Basic hypotheses
In accordance with the above scheme and discussion, the following hypotheses can be suggested:
(a) Pollution deposition and transport are two distinct phenomena, the first being chiefly connected with human activity and land use, the second being connected with rainfall and drainage area characteristics.
(b) The total accumulated mass of a pollutant over a long period (a year for instance) may be quite constant (or with a 'sharp' normal distribution) for a given basin.
(c) The annually transported mass of a pollutant is sensiby constant and equal to the accumulated mass (if the mass of pollutant is conserved).
(d) Pollution deposition is a stationary phenomenon.
(e) Land use is the major influence on pollutant accumulation.
(f) For a given land use and a given pollutant, the accumulated and transported masses have a linear variation with the area (no scale effect).
32 Michel Desbordes étal.
Possible studies
These hypotheses lead to some possible studies related to the two research objectives defined earlier. Consider a number of land use groups for a year-long measurement period, during which a mean sample after each rainfall event is obtained, the annual transported mass, MEik of a pollutant k for the /th basin may be :
2
N m,k x St] (3)S if is the land use area of the/th type in the /th basin. If the hypotheses (c) and (f) above are satisfied, mJk could be described as an annual specific mass of pollutant k for land use /. However m/fc must not be related to the basin and climate character- istics (low air pollution, etc.). mJk may be obtained if all the rainfall events are sampled, if there are at least N experimental basins, and if each land use exists at least once. A greater number of basins will of course be desirable in order to study a possible connection of mik with the basin climate characteristics. The basins must preferably be located in different geographical areas, and there should not be an unusual climatic event during the observation period (such as a very long dry spell).
The determination of m/fc allows the comparison of:
— the mass of a pollutant k for/ different land use types:
Rk=mak/mbk, j = a,b
— the mass of A: different pollutants for the same land use/:
Rf = mja/mfb, k = a,b
These studies may satisfy the first research objective. For the second, the time scale must be smaËer (the duration of the rainfall event for example). In fact, it is the study of the rate of change, and its effect on the quality of the receiving water that would determine the time scale. At present, there is very little information on this particular subject, and more research is required. However, it seems necessary to use paired basins with the same sequence of rainfall events. (This is necessary anyhow for the case where mass of pollutants is not conserved.) The underlying system may then be solved:
MEipk
2
N _mipkxSif (4)with at least N basins in the same urban area. If only one urban area was used, it would not be possible to study the climatic influence, or the interdependence between the deposition and transformation phenomena, etc. The study of the distribution of the (Cj ) ratios such as
RPk = mapk /mbpk, j = a,b
may provide interesting information about the phenomena. For example if these ratios were normally distributed, with a mean value close to the Rk annual values, the following hypotheses may be advanced :
(1) The mass of pollutants is conserved.
(2) The deposition laws for different land uses are significantly homothetic.
(3) Many possibilities can be found for the transport laws.
Fither (i) almost all the mass of pollutant is removed from the basin during each rain- fall event ; or (ii) the transport phenomenon is independent of the land use, and the transported mass is always proportional to the available mass over the basin; or (iii)
Urban runoff pollution in France 33 the transport laws are homothetic from one land use to another and in the same ratio as the accumulation laws. However, other conditions may lead to closely constant Rpk ratios for special experimental cases (dry weather durations with little change around the mean value, homothetic transportation laws).
Moreover, if the Rpk values were not normally distributed around the Rk annual values, nothing could be said about the homothesis of the accumulation laws or about both (accumulation and transportation) laws together. For example if the mass of pollutants is not conserved, the Rpk ratios would be constant in only very special experimental cases (very short mean dry weather duration, etc.).
For the modelling of A (accumulation) and E (transportation) phenomena, there are several alternatives:
(1) Make two conceptual models ÎQIA and E phenomena (another one for R, the residual pollutant, if the mass pollutant is not conserved); then try to fit them to the data. But the models can be used on ungauged basins only if the fitted parameters could be defined in terms of basin characteristics and/or land use.
(2) Make only one conceptual model for A or E (two if the mass of pollutants is not conserved); then look for a statistical model for the second (third) phenomenon.
For example, a model giving the Ap values may be developed (or Ap and i?p_ j together) then the E laws may be derived statistically. So for a pollutant k and a land use; (on the rth basin) the following system must be solved:
EjPk = Ajpk +RJ\p-i-k -Rjpk (5)
where Ajpk is a conceptual model ; Rj p _ t k is a conceptual model (k non conserved) ; and
R;pk = / (rainfall characteristics, Alpk, Rj p _ t > k, etc.) (6) This system will give a solution where the goodness of fit will depend upon the model
used for A (and R ). In this case, using the total transported masses, it is not necessary to study the time distribution of the E phenomenon and so the number of expensive analyses may be considerably reduced. On the other hand, if an E model is used, it would be desirable to fit that model to data and then to build 'pollutographs'. The reconstruction of such a 'pollutograph', for a pollutant k, coming from an area under land use / on the rth basin (equations (5) and (6)) may be done using a runoff model and a simple pollution routing model. Now that the E model has been constructed, a statistical or conceptual A (and R) model may be developed using trial and error techniques. At the present time, it is not possible to say which of these methods would be preferable.
Finally, if there exists a scale factor for the phenomena, equations (3) and (4) may be written as
2
N Cjk Vif (7)2
N Cjpk Viip (8)For a given land use/, the 'concentration' terms are C/fe and C/pk. The Vtj and Fyp
runoff volumes may be given by a runoff model. The 'concentration' terms can then be used as the above 'specific masses'.
34 Michel Desbordes et al.
Minimum experimental design
Considering the research objectives, the constraints, the hypotheses and the possible studies discussed above, the following minimum experimental design requirements can be identified:
(1) Find basins consistent with the measurement constraints. This will lead to a careful choice of basins and important preliminary studies.
(2) For the N land use types studied, choose at least N basins in the same urban area.
(3) Choose at least N other basins in a different area.
(4) Take a mean sample for all the rainfall events over the observation period (which will be at least a year long).
(5) Take some discrete samples (pollutographs) for some events (dry weather- rainfall pairs). It is not yet possible to give the number and type of events that should be sampled. It only can be hoped that, in the first instance, these events may produce important deposited and transported pollution in order to minimize the measurement errors: long dry spells, extreme intensity or volume of rainfall, etc.
METHODS OF SAMPLING AND MEASUREMENT
Following the experience in other countries, an automatic sampling device was chosen. The representativity problems of the samples (strainer's position, ratio of sampling, etc.) have been discussed by the working group. Some EPA reports (Shelley and Kirkpatrick, 1973; Shelley, 1975), a French study (Cathelain, 1979) and a few estimates of the influence of strainer position have also been used. Finally it was decided that the automatic sampler must have a high-rate peristaltic pump with a strainer having always the same position in the flow.
A sampling and measurement system with limited complexity has been developed to satisfy the experimental design considerations. The availability of funds has also been an important element in the choice, but it was preferred to reduce the number of measuring points rather than to use bad quality data.
Finally the sampling and measurement system for each experimental basin will include: (1) two tipping bucket raingauges with a 1000 cm2 receiving area, (2) two automatic samplers (with samples proportional to the discharge), and (3) a water level measurement device which starts the system for a given flood, continuously measures and records the water level in the sewer, and controls the frequency of samples. The stage—discharge curve, for a given sewer will be obtained using an hydraulic control (weir, venturi flume, etc.) or by the use of radioactive or chemical tracers. One of the samplers will give a mean sample for the whole duration of a rainstorm, the other will give discrete samples for use in constructing pollutographs.
The choice of which pollutants to analyse is also a compromise between the money available, the researchers' experience with respect to pollutants usually examined (BOD5, COD, etc.), and the micropollutants themselves and their impact on the receiving waters used for human consumption. For each discrete sample the following analyses will be undertaken: BODs before and after 2 h sedimentation, COD before and after 2 h sedimentation, and suspended solids.
But, for the mean sample, the following analyses will also be undertaken: NKjeldahl, NHj, N 03, PO|~, phenol, non-floating total Pb, Zn, Cd, Ni, total Cr, Mn and Mg.
Some local analyses of rainfall pollution and some sieve analysis curves of suspended solids will also be done.
In order to satisfy the experimental design considerations, the measuring system must be kept in good condition for the duration of the experimental programme, i.e.
one year. For this purpose, it has been decided that the operational field unit will
Urban runoff pollution in France 35 collect the samples after each rainfall event, take care of the measuring system three times a week, and undertake the re-synchronization of the raingauges and water level recorders.
The supervision of the overall programme will have three stages:
(1) The operational field units will make a first appraisal of the results.
(2) A second appraisal will be made for all the basins by the members of sub-group (2). Then the results will be coded and digitized.
(3) A preliminary analysis will be undertaken by a research group. This group will give information with respect to how successful the programme has been, what type of rainfall event needs to be sampled and analysed, etc.
TABLE 1. Choice criteria
Criteria
Dry weather flow as percentage of maximum flow in the sewer
Urbanization homogeneity
Possibility of local grants
Climatological data
Measurement devices and data
Site accessibility Slope of the basin
Experience of the measurement team
Quality of the answer
Sewerage system condition Street sweeping?
Building sites?
Measurement easiness?
Bibliography?
Note
Waste water flow Waste water flow Seepage
Seepage Number of areas Number of areas Number of areas Number of areas Number of areas Nothing
<50 000 F
> 5 0 0 0 0 F \
< 1 0 0 0 0 0 F )
> 100 000 F Average Good Rainfall Flow Water quality Difficult access Easy access
<0.5%
> 0.5% < 3%
> 3 % Poor Average Good Poor Average Good Poor Good Yes, unknown Yes, known No Yes No Difficult Average Easy No Yes
Very good one
>5%
<5%
>5%
<5%
> 5
= 4
= 3
= 2
= 1
l o i l o i l o i
1) 2 3
4J
1\
2 3 2 I)
il
4) 3
1\
2\
~2\
:21) \
I)
2 3 1i) n
i 2 3 1) I)
Ù w
1)
i)
UÛ
Weighted factor
X5
X5
X5
X 4 X 4 X4
X 3
X 3 X 2 X 2 X 2
X I
36 Michel Desbordes et ai.
TABLE 2. Selected basins Name of the basin
Beauvais Créteil Les Ulis
Plaisir La Boissiere Aix - Z.U.P.
Maurepas Plaisir Centre Aix Nord Clichy sous Bois
Climatological area
1 1 1 1 3 1 1 3 1
Surface area [ha]
43 46 49 31 30 26 247 105 72
Imperviousness coefficient 0.54 0.53 0.42 0.31 0.90 0.60 0.29 0.45 0.43
Mean slope l%]
1.58 4.00 0.55 5.00 3.00 0.50 0.90 6.50 2.00
CHOICE OF EXPERIMENTAL BASINS
Sub-group (1) has been working in two directions with respect to this choice. The first one was in the definition of the criteria in relation to the experimental design, the literature review giving the most important parameters in the runoff pollution study, and the constraints of the measurement method.
The second direction led to a questionnaire, the answers to which could influence the final choice. A similar questionnaire had already been circulated during the first experimental programme for the study of urban runoff quantity. A simple one was first sent to cities having more than 20 000 people, to local drainage community associations, to private firms, etc. About 70 useful answers were received (3000 questionnaires sent!). The basins were then classified as shown in Table 1 using weighted factors for the parameters. From this study, nine basins were selected (Table 2). Comprehensive questionnaires were then sent to these nine in order to make a final choice. Finally the whole working group have made a last classification among these basins.
The measurement programme will probably begin in 1980 for four basins: Maurepas and Plaisir La Boissiere in the Paris area, and two located in Aix-en-Provence in the south of France. Each pair can be assumed to have homogeneous rainfall events. The land use type is single-family housing and new residential multi-family housing.
CONCLUSIONS
Due to the relatively late start, the study of runoff pollution in France is benefiting from the experience in other countries. It is recognized that the measurement of runoff pollution is a difficult and expensive problem. It is hoped that the first national measurement programme will give some information about the validity of the experi- mental design even if it does not completely satisfy the two objectives described. An international comparison of results and if possible, international cooperation would probably accelerate a complete understanding of these problems and limit the number of expensive programmes which often lead to quite poor results.
REFERENCES
Cathelain, M. (1972) Investigation on automatic server flow samples. Service Technique de l'Urban- isme, September 1979.
Desbordes, M. and Ribstein, P. (1978) Study of urban runoff quality: a literature review synthesis.
LHMReport no. 45, Montpellier University, October 1978.
Shelley, P. E. (1975) Design and testing of a prototype automatic server sampling system. US EPA Report no. R 2-75, February 1975.
Shelley, P. E. and Kirkpatrick, G. A. (1973) An assessment of automatic server flow samplers. US EPA Report no. R 2-73-261, June 1973.
