The Science of the Total Environment, 33 (1984) 213-224 63

Elsevier Science PublishersB.V., Amsterdam-Printedin The Netherlands

SEDIMENT SUPPLY AND TRANSMISSION VIA ROADSIDE GULLY POTS

C.J. PRATT' and J.R.W. ADAMS'

1 Department of Civil and Structural Engineering,
Trent Polytechnic, Nottingham, NGl 4BU, (United Kingdom)
2 Pollution Control'Dept., Boots PLC., Nottingham, (United Kin.gdom)

ABSTRACT
Field studies undertaken on the Clifton Grove residential estate,
Nottingham, between August 1976 and October 1980, in which the
input and release of material to and from roadside gullies and the
gully pot liquor quality were monitored, are reported. Factors
affecting sediment supply, such as variations in soil moisture.
deficit, rainfall parameters, runoff volume and catchment character-
istics are considered, Sediment retention by gullies and,its
release in storm flows are assessed in the light of the nature of
the sediments found in basal samples from gully pots and in runoff
sample% obtained downstream in the storm sewer. The effect of
sediment input on gully liquor quality and the variation in
dissolved oxygen concentration are detailed. The suspended sediment
concentration in road surface/gully pot discharges-showed no clear
tendency to decrease through a storm event and this, coupled with
previous findings from storm simulations, strongly suggest that
"first flush" effects are a function of the flushing of in-pipe'
deposits from a previous storm event.

INTRODUCTION
Urban storm drainage research has been conducted,on the Clifton
Grove catchment, Nottingham, since 1976 and,in that time aspects of
both the quality and quantity of stormwater runoff have been
investigated (ref.l-5). The catchment has a total area of 10.6ha'
of which 42% are impervious highway and roof surfaces, draining to
a separately-sewered storm drainage system. The roads, pavements
and driveways, which drain to the sewer system, represent, elmast
half of the total impervious surface area.
Built between 1973 and 1976, the Estate is spaciously arranged;\,
with areas of open land interspersing the rows of housing .along,gide 1
cul-de-sacs and the loop, .feeder road (ref.133). ,There arme.266
detached houses and bungalows and an estimated population. of 9:G2,
assuming one person per bedroom. All the properties have front
and' rear gardens, the front gardens generally being open without

00489697/84/$03.00 81984Eleevie.r.ScienoePubliahersB.V.
214

walls or hedges separating them from the footway: gsrage and drive-
ways allow at least two vehicles to be accommodated per household.
The highway, surface water drainage is via gratings in the kerb-
2
side channels, each grating draining on average 185m of footway
and carriageway, which are both surfaced in tarmacadam. Pr evi 0u s
research on the overland flow aspect of stormwater runoff on this
catchment (ref.4) has allowed estimates to be made of storm runofE
inlet hydrographs to gullies, being studied for water quality
purposes, without the interference of flow monitoring equipment.
The paper reports the findings of two field studies in which the
input of material to roadside gullies, gully liqour quality and the
water quality of road, roof and total catchment discharges were
monitored. Although no two catchments are identical, the results
obtained may be typical of other modern residential estates,
composed of middle-income housing, which are some distance from
industrial or commercial activity.

SEDIMENT SUPPLY TO GULLIES

Five gully catchments were monitored between August 1979 and
October 1980 (see Table 1). I\laterial entering the gullies was
collected in a plastic box, sealed in position just below the
grating, and on one of five conical nylon meshes, which received
flows from the box prior to discharge into the gully pot below (ref

3). The meshes had apertures of 1250, 600, 400, 150 and 90 microns
It was hoped that the use of several meshes, retaining different
fractions of the material washoft, would avoid clogging leading to
overflow, which seemed the case in all but one storm event.
The collecting system was cleaned and the meshes changed after
14 days. On one occasion a grating was frozen into its frame and a
28-day sample was collected. At each of the other gullies, data
from one 14-day period was lost due to vandalism, otherwise the
collecting system worked satisfactorily to October 1980: for gullie
6 and 34 from August 1979; gully 28A from November 1’979; gully 52
from January 1980; and gully 84 from October 1979.
The mass of material washotf collected in a 14-day period was
divided into two components: organic debris, being material larger
than 1250 microns comprising leaf litter, flower petals, grass
cuttings and paper; and total sediments, which were the sum of the
masses of organic and inorganic sediments retained on the meshes.
 215

@rganic debris
The supply of material and its nature were variable throughout
the year and across the catchment. Some organic debris was water-
borne into the gullies, but much was wind blown and need not have
originated in the particular gully catchment being monitored.
Seasonal variations were clearly evident e.g. autumnal leaf-fall
between October and December and the summer shedding of flower
petals and grass cutting debris from late Nay to September. The
period of limited plant matter supply and restricted gardening
activity, Erom February to April, presumably accounted for the
generally low levels of organic debris input to the gullies during
that period (ref.3).

Organic an.d inorganic sediments

Whilst sediments, as the organic debris, may be readily moved by
wind and vehicle-generated forces, their predominant mode of
transport to the gully inlets is believed to be within the storm-
water runoff from the highway surfaces. The action of the kerb face
as a barrier against which sediments accummulate has been reported
(ref.6). Once located in the kerbside channel, the sediments lie
in the path of surface water streams to be transported in suspension
or by rolling to a gully during storm events. However, there was
found to be generally no correlation between the 14-day mass of
sediments washed from the highway surface and the runoff volume
within the period (ref.3). It is clear that the sediment washoff
is dependent more upon the factors influencing sediment supply to
the highway surface and its mobilisation than upon the total
discharge which transports it to the gully.
Only when the soil moisture deficit fell to zero, between late
December and early April, was there any correlation, and that not
high, between'the 14-day mass of sediment vashoff, SED, and the
runoff volume in the period, V (ref.3).
Gully 6, SED = 0.005V t 9.5 (r = 0.37) (1)
Gully 34, SED = 0.015V t 16.7 (r = 0.63) (2)
For the five monitored gullies,
Average, SED = 0.017V (r = 0.63) (3)
Between December and April, the mass of sediment vashoff was
generally low e.g. gully 6, SED = 3Og, V = 3200 litres; gully 34,
SED = 2oog, V = 4800 litres. The binding effect of the surface
moisture in the surrounding areas limited the dispersion of
particles onto the highway by natural forces. Limited gardening
216

activity and pedestrian movement would also play a part in restrict-

ing the dispersion of sediments at that time of year. From April
to December no correlation esisted between sediment washoff and the
runoff volume, and the variations in the mass of sediment washoff
at similar runoff volume were sometimes considerable e.g. gully 34,
SED 40 - 4205 for 1' about 2800 litres (ref.3). Predicting sediment
washoff becomes extremely complex, as the various interactions of
catchment characteristics, meteorological conditions, hydrology
and hydraulics influence the supply and transport of sediments.
One 14-day period (lo-24th April, 1980) was without rainfall.
The masses of sediments collected in gullies 6, 28A, 34 and 52,
between 4000 and 90 microns, were almost identical despite the
differences between the catchments: mean mass collected by the
meshes was 3.61 !: 0.066g. llad the soil moisture deficit been other
than near zero, these similar results would seem unlikely to have
occurred.

Catchment characteristics
Table 2 details the median and mean 14-day masses of sediment
washoff for the five monitored gullies for 19 and 26 observation
periods. The long-term mean values of washoff, S!?n (g), were well
correlated with the gully catchment characteristics, namely,
impervious area, IMPAREA, maximum drainage path length, Lmax, mean
catchment slope on the maximum drainage path length, SLmax and the
number of houses served in and via the catchment, HOUSES, assuming
the shortest route taken to each house from the Estate entrance.
SEij = 1.6 IMPAREA - 151.6 (r = 0.99) (4)
---
SED = 12.4 LmaY - 283.2 (r = 0.91) (5)

SEij = 45.9 SLmav - 30.8 (r = 0.92) (6)

r?Eb = 1.4 HOUSES + 39.8 (r = 0.86) (7)

NattIre of sediments
Examination of the mesh-trapped sediments revealed that the
solids retained on the 1250, 600 and 400 microns meshes, which
represented about 92% of the total mass (1250 - 90 microns),
contained only 4% of the volatile solids. The remainder trapped on
the 150 and 90 microns meshes represented, repectively, 6% of the
total with 29% volatile and 2%, but with 90% volatile. The total
volatile fraction was 7% of the total sediment mass. This
distribution of sediment sizes and their nature has important
 217

TABLE 1
Catchment characteristics of gullies monitored on Clifton Grove.

Gu 1 1 y Impervigus Slope Length Number of

location area (m ) (%) Cm) houses served
n u mb e r I HP i\ R E A S Lmax L in and via
ma s
43a 140 4.2 29 36
53" 290 4.3 3s 6
SGa 313 4.0 36 S6
465 2.9 4s 11
103 1.4 26 11
111 0.7 23 21
248 4.9 38 153
176 4.3 36 11
213 4.4 40 83

a(ref. 1 R 7), August 1976 - June 1977.

b(ret. 3 R S), August 1979 - October 1980.

TABLE 2
Mean 14-day masses of sediment washoff to roadside gullies on
Clifton Grove, Nottingham.

Gully 14-day sediment washoff into gully (g)

location lledian of Elean of Elean of
number 19 periods 19 periods 26 periods
Jan. 3rd - Oct. 9th 1980 Oct. 4th 1979
- Oct. 9th 1980
6 19 25.7 25.4
28A 4.3 11.1
34 110 249.2 215.4
52 57 130.2
84 73 172.9

bearing upon the retention characteristics of gully pots.

SEDIblENT RETENTION BY GULLY POTS

An estimate of the mean composition of the basal sediments in
the Clifton residential area of Nottingham was obtained from a
sample taken from load collected by a gully emptying machine (ref.7
see Table 3). The bulk density of the basal sedinents in gully pots
on the Clifton Grove catchment was 1492kg/m3 (c.f. sands, 1770kg/m3)
with mean particle sieve diameter 1.75mm. Mean particle diameter
of mesh-trapped sediments for gullies (in Table 2) was 0.7mm.
Comparing the gradings.of the basal and the mesh-trapped sediments
(see Table 4) shows that, assuming the gully pot is reasonably
218

TABLE 3
Composition of basal sediments from gully pots in the Clifton
*rea of Nottingham.

Insoluble components: Volatile solids (2) 21.0

DOD (mg/g) 43.5
COD (mg/g) 34711
Soluble components: Dissolved solids(mg/g) 4.0
COD (mn/E) 3.2

TABLE 4
Comparison of sediment size fractions in gully basal deposits and
14-day washoff into roadside gullies.

Sieve Aperture Mass retained on sieve Cram llcg sample

(microns)
Rasal sediments Mesh-trapped sediments
(9) in 14 days washoff (g)
5000 140 60
2000 280 60
1000 200 180
500 160 500
300 110 140
300 110 60

TABLE 5
Annual mean concentrations of constituents for individual eullv
pots in mg/l for contained liquor.,

Constituent Gullv location number

d d d
43= 53= 86c 97‘ gd 2Sd 33 35 50d 86
Suspended solids 15 15 29 27 27 26 19 20 24 24
Dissolved solids 197 818 231 343 191 269 163 209 140 162
COD 25 30 37 56 39 109 54 68 57 47
ROD 7.1 6.5 6.5 7.1
Dissolved oxygen 3.6 6.0 6.2 8.5
Pll 7.5 7.9 7.5 10.5 7.8

'(reK. 7), August 1976 - June 1977.

d(ref. 8), August 1979 - October 1980.

effective at retaining 2mm sediments, four to five times the mass

of sediment input would be required to accummulate the observed
mass of 2mn sediments retained. This suggests that the mass of
sediment wash-through exceeds the mass retained by many times for
sediments less than lmm sieve diameter.
 219

G II I, I, I’ I, T Q I IO R (,I I1 A I, I ‘I’ 1
Gully liquor quality has been found to vary, not only seasonally,
but across the Clilton Grove catchment between ditferent gullies
(ref.147). Ta b I e s 5 a nd 6 di sp1Ry results obtained between 1976
and 19SO.

Effect of sediment intercention

Comparison was made between the gully liquor quality of meshed
gullies (location no.5 6, 28A, 34, 52 R 84) and of adjacent,
unmeshed gullies with simil.nr catchments (5, 25, 33, 50, 86) for
the period January Lo October 19SO (ref.8). Considering mean values
for the two groups ol gullies in each 14-day period sample, COD was
consistently lover by 35 + 3;% in the grollp of‘ meshed pots. On one
occasion the mean suspended solids concentration in the meshed pots
just exceeded that in the unmeshed: dissolved solids concentrations,
al though not necessarily lover in the meshed pots, were so in about
two-thirds of the samples.
Without simultaneously monitoring the inElows and outflows from
a gully pot, it is not possible to determine accurately the
efficiency ot a gully at trapping sediments of various sizes and
types. Laboratory experiments have shown that with 50mm oE gully
basal sediments (as Table 3) and flow rates up to 1 l/s, less than
f% of the basal sediments were re-suspended and washed out of the
pot (reE.1); and using sediment increments of artificially
constituted sand grading, typical of observed city street sediments,
on the rising limb of a 20-minute triangular hydrograph, with peak
discharge 1 l/s, some 9% by weight of sediments between 300 and 150
microns washed directly through the pot and 11% for sizes between
150 and 90 microns (reE.ll). If the latter experiment was repeated
with a grading of sediments composed of the volatile Eraction, it
seems likely that the percentage wash-through in the 300 - 90 micron
would be several times greater than observed, in view of the diff-
erence in specific gravities of quartz sand grains, as tested, and
the volatile fraction. This aspect and others, associated with the
performance of gully pots in determining stormwater discharge
quality, will be investigated at Trent Polytechnic in research
commencing in September 1983, funded by the Science and Engineering
Research Council, Swindon.

Variations in dissolved oxygen concentration

The dissolved oxygen concentration of the gully pot liquor may
TABLE 6
Mean quality of water stored in gully pots on Clifton Grove (mg/l)
from research between August 1976 to June 1977 (ref.7)

Constituent Number of Mean Maximum

samples
Suspended solids 243 31.2 455
Dissolved solids 261 335 17475
COD 176 63 935
BOD 195 7.6 135
Dissolved oxygen 63 6.0 0 - 11.2
Ammoniacal nitrogen 266 1.0 5.9
Nitrate nitrogen 263 2.0 29.8
Chloride 220 268 9800
PH 201 8.0 6.3 - 12.9
Calcium 246 31 272
Sodium 221 125 5560
Potassium 221 3.8 28
Lead (total) 31 0.058 1.36
Lead (soluble) 11 0.032 0.10
Zinc (total) 74 0.65 3.5
Zinc (soluble) 29 0.10 0.56
Copper (total) 28 0.012 0.32
Copper (soluble) 7 0.05 0.09
Cadmium 26 0.058 0.07
Manganese 19 0.21 0.74
Nickel 19 0.07 0.42
Iron 19 1.08 7.00
Chromium 19 0.05 0.48

be a useful indicator of the liquor's condition and of the level

and type of bacterial activity. The concentration to achieve
saturation decreases with increasing temperature, hence the
recharge of dissolved oxygen during storm runoff will vary according
to daily temperature and the season of the year. Measurements of
gully liquor maximum and minimum temperatures showed that they fell
closely about the mean of the daily maximum and minimum air temper-
atures, but that occasional major variations might occur, which
could be attributed to runoff from surfaces made very. hot by the
sun or from snowmelt e.g. 29.5' and 2.5'C, respectively (ref.8).
Generally, the DO concentration was 60% of the saturation value at
liquor temperatures between 2.5' and 20°C (i.e. 6.5 to 3.5mg/l),
but only 25% of saturation levels at temperatures around 25'C (i.e.
at 1.2mg/l), for some unknown reason.
Mean values of DO concentration for gully liquors (as in Tables
5 & 6) are no indicator of daily liquor condition, but they may
offer some guidance as to the gully catchment areas necessary, in
the long-term, to provide adequate runoff volume to effect a
 221

measure of oxygen recharge generally. For gullies 43, 53, 86 and

97, mean gully liquor DO concentration could be related to the
impermeable catchment areas:
ijo = 0.015 IMPAREA t 1.5 mg/l (r = 0.99) (8)
For Clifton Grove, the average gully catchment area is 185m*. Mean
DO concentration would be estimated to be 4.2mg/l (Eq.8), which
corresponds to about the mean observed concentration in gullies 6,
28A, 34, 52 and 84 at a liquor temperature of 15'C, during a summer
period with high liqour pollutants potential. To raise the mean DO
concentration to 6mg/l would require impermeable gully catchments
of 300m2 (Eq.8). A previous estimate of the required gully catch-
ment area to achieve this same mean DO level, based upon mean
summer weekly raintall totals (ref.l), suggested areas up to 500m*
were necessary. Certainly, catchments considerably larger than the
present 200m* design recommendation (ref.9) seem in order, if
maintenance of gully liquor quality is being considered as an
important criterion at the design stage of a residential development
The influence of the antecedent dry period duration on suspended
and dissolved solids concentrations in the gully liquor, of human
activity in the gully catchment on liquor quality and the relation-
ships between COD and BOD and catchment characteristics have been
reported previously (ref.1).

SEDIMENT TRANSFER FROM GULLY CATCHMENT TO SEWER SYSTEM

Samples of gully pot discharges were obtained at three locations
(gullies 1 C 2, 35, 46) on Clifton Grove during 32 runoff events,
with sampling duration from 10 to 150 minutes, to assess the
contribution of the road surface and gully liquor pollutants to the
total catchment discharge quality. Tables 7 and 8 summarise mean
quality parameters for road surface/gully, roof surface and total
catchment discharges, unfortunately, from different research periods
For the road surface/gully pot discharges, in only 18% of the
catchment events was the final suspended sediment concentration
less than half the initial sample's concentration, and in 44% the
final concentration was actually higher than the initial value e.g.
gully 35 for 24 samples obtained during a storm on 28th October 1980
initial sample suspended solids concentration 77mg/l, final sample
concentration 95mg/l, mean 83mg/l, range 35 - 165mg/l, mean volatile
suspended solids 26mg/l. This would suggest that there was no
obvious exhaustion of.sediments at source, an observation which has
also been made by Reinertsen .(ref.lO). The "first flush" and
222

TABLE 7
Mean pollutant concentraLions in road surface/gully pal. and roof
discharges and in atmospher-ic fallout for Clifton Grove, Nottingham
August 1979 - October 1980 (ref.8)

Constituent Road surCace Roof surface r\Lmospheric

/gully pot discharge raIlout
discharge
(ms/l) (mg/l) (ms/l)

Suspended solids 96.6 12.3 24.0

(7 - 608) (1 - 83) (6 - 72)
Dissolved solids 157.5 202 71.4
(16 - 1986) (78 - 740) (20 - 243)
Total solids 249 216 97.3
(36 - 2070) (88 - 762) (21 - 268)
COD 113.5 57.9 43.5
(4 - 615) (8 - 360) (5 - 1SO)

TABLE 8
Mean pollutant concenLrations in total catch discharge for Clifton
Grove, Nottingham, August 1977 - June 1977 (ref.7)

Constituent Mean Maximum Number of

(ms/l) (ms/l) samples

Suspended solids 94.5 852 257

Volatile suspended
solids 26.8 60 22
Dissolved solids 119 492 183
DOD 8.7 44.5 125
COD 63.4 600 147

"exhaustion of sediments" apparent in total catchment discharge

monitoring may be explained as functions of in-pipe deposition and
the relative times and quantities of runoff from road and roof
surfaces.
A storm runoff simulation model was developed for the Clifton
Grove catchment (ref.5) and used to predict the component hydrograph
at the outfall due to road and to roof surface runoffs across the
catchment. Whilst each storm produces different hydrographs and
their relative magnitudes and timings differ, it is generally the
case that roof runoff preceeds road surface runoff, contributes the
earlier and higher peak discharges and is first to cease in the
recession limb. Therefore, iE suspended sediment input from road
surfaces via the gully pots is not exhausted during a storm event;
 223

roof surface-originating suspended solids concentrations are consid-

erably I.over than those from road surfaces (see ‘Table 7); and
“first Ii ush” effects occur during predominantly roof-only runofr,
the initial, high suspended solids concentrations in the outfall
discharges seem likely to be the result of’ the flushing of pipe
deposits, originating on the recession limb of a previous storm
event.
Mean observed pollutant concentrations in stormwater discharges
at the outfall from the Clifton Grove catchment are given in
Table 8.

Factors influencing sediment transfer

The dominating effect of total rainfall, or total volume
discharged, in determining the pollutant loading of flows in storm
sewers (ref .l’Z) may mask the influence of other meteorological
parameters and catchment characteristics. Multiple regression
analysis, using data from 19 events on the three road surtace/gully
pot catchments indicated that rainfall intensity parameters, peak
1-mi.nute, 5-minute and mean intensity, were the major factors
influencing the flow-weighted, mean suspended solids concentration
of a runoff event, explaining 90% of the variance. Only 3% of the
variance was associated with catchment slope and 0.0% with the
duration of the antecedent dry period. This supports the conclusion
that sediments are not necessarily exhausted from the surface of the
h i g h !r’ a y , suggesting that it is factors affecting sediment mobilis-
ation which are most important, not the period available for the
accummulation of sediments before the runoff event.
For COD and dissolved solids, the antecedent dry period was of
greater importance, explaining 15 - 90% of the variance for the
individual gully catchments.

CONCLUSIONS
A summary of field research findings, concerning sediment supply
and stormwater runof E quality aspects, have been presented for a
typical middle-income, residential area in the U.K., monitored
between August 1976 and October 1980. ‘The performance of the gully
pots within the storm drainage system, as interceptors, retention
devices and as sources of sediments and associated pollutants has
been examined. Orders of magnitude of some pollutants discharged
from a residential area have been reported and the need for further
detailed study of pollutant discharges from such catchments is
224

recommended. In total area across the U.K., residential catchments

represent a very significant source ol stormwater runoff and
associated highway pollutants discharge, the latter being similar
in some constituent concentrations CO those monitored on motorways
and in commercial areas.

REFERENCES

1 I.J. Fletcher, C.J. Pratt and G.E.P. Elliott, An assessment of

the importance of roadside gully pots in determining the quality
of stormwater runoff, in P.R. Hellivell (Ed.), Urban Storm
Drainage, Pentech Press, London, 1978, pp. 586-602.
2 I.J. Fletcher and C.J. Pratt, Mathematical simulation of pollut-
ant contributions to urban runoff from roadside gully pots, in
B.C. Yen (Ed.), Stormwater Quality, Management and Planning,
Water Resources Publications, Littleton, U.S.:\., 1982, pp.116-124
3 C.J. Pratt and J.R.W. Adams, Sediment washoff into roadside
gullies, in B.C. Yen (Ed.), Stormwater Quality, Management and
Planning, 1v'ater Resources Publications, Littleton, U.S.A., 1982,
174-183.
4 K . Pratt and R.J. Henderson, Overland flow and catchment
characteristics, in B.C. Yen (Ed.), Urban Stormwater Hydraulics
and Hydrology, Water Resources Publications, Littleton, U.S.A.,
1982, pp. 83-92.
5 C.J. Pratt and J.J. Harrison, Storm runoff simulation on a
calibrated catchment, in R.E. Featherstone and A. James (Eds.),
Urban Drainage Systems, Pitman, London, 1982, pp. 2-143 - 2-155.
6 J.B. Ellis, The nature and sources of urban sediments and their
relation to water quality, in G.E. Hollis (Ed.), Mans Impact on
the Hydrological Cycle in the U.K., Geo Abstracts, Norwich, 1979,
199-210.
7 T"; . Fletcher, Quality of urban stormwater runoff: an appraisal
of the role of gully poy liquors, Unpublished PhD Thesis, Trent
Polytechnic, Nottingham, 1981.
8 J.R.W. Adams, Contaminants in urban stormwater, Unpublished PhD
Thesis, Trent Polytechnic, Nottingham, 1983.
9 Anon., Guide to the requirements for the design of highways and
paved areas in residential and industrial development,
Nottinghamshire County Council, Dept. of Planning and
Transportation, Nottingham, 1978.
10 T.R. Reinertsen, Quality of stormwater runoff from streets, in
B.C. Yen (Ed.), Urban Stormwater Quality, Management and Planning
Water Resources Publications, Littleton, U.S.A., 1982, pp. 107-
115.
11 M.C. Wilson and C.J. Pratt, Unpublished laboratory results,
Trent Polytechnic, Nottingham, 1979-1982.
12 G. Mance and M.M.I. Harman, The quality of urban stormwater
runoff, in P.R. Hellivell (Ed.), Urban Storm Drainage, Pentech
Press, London, 1978, pp. 603-617.