See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/233613081

The role of inter-event time definition and recovery of initial/depression loss for
the accuracy in quantitative simulations of highway runoff

Article in Urban Water Journal · March 2007

DOI: 10.1080/15730620601145873

CITATIONS READS

16 89

4 authors, including:

Rupak Aryal Hiroaki Furumai

University of South Australia The University of Tokyo
90 PUBLICATIONS 885 CITATIONS 280 PUBLICATIONS 2,586 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Analysis of pathogenic distribution during inundation using model XP-SWMM for the old city of Hue View project

Permeable pavement clogging and stormwater View project

All content following this page was uploaded by Rupak Aryal on 22 July 2015.

The user has requested enhancement of the downloaded file.

This article was downloaded by:[Tokodai Kinokunya Co]
[Tokodai Kinokunya Co]
On: 26 February 2007
Access Details: [subscription number 759158443]
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954
Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Urban Water Journal

Publication details, including instructions for authors and subscription information:
http://www.informaworld.com/smpp/title~content=t713734575
The role of inter-event time definition and recovery of
initial/depression loss for the accuracy in quantitative
simulations of highway runoff
To link to this article: DOI: 10.1080/15730620601145873
URL: http://dx.doi.org/10.1080/15730620601145873

Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf

This article maybe used for research, teaching and private study purposes. Any substantial or systematic reproduction,
re-distribution, re-selling, loan or sub-licensing, systematic supply or distribution in any form to anyone is expressly
forbidden.
The publisher does not give any warranty express or implied or make any representation that the contents will be
complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should be
independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings,
demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or
arising out of the use of this material.
© Taylor and Francis 2007
 Urban Water Journal, Vol. 4, No. 1, March 2007, 53 – 58

The role of inter-event time definition and recovery

of initial/depression loss for the accuracy in quantitative
simulations of highway runoff
Downloaded By: [Tokodai Kinokunya Co] At: 03:59 26 February 2007

R. K. ARYAL{, H. FURUMAI*{, F. NAKAJIMAx and HKPK JINADASA{

{Department of Environmental Science and Engineering, Kathmandu University, Dhulikhel, Kavre, Nepal
{Department of Urban Engineering, University of Tokyo, Hongo 7-3-1, Bunkyo-ku, 113-8656 Tokyo, Japan
xEnvironmental Research Center for Advanced Science and Technology (RCAST), University of Tokyo,
Hongo 7-3-1, Bunkyo-ku, 113-0033 Tokyo, Japan
{National Water supply and Drainage Board, New Road, Bandarawela, Sri Lanka

A long-term runoff monitoring was carried out in a highway drainage system in

Winterthur, Switzerland. Several runoff quantity simulations were carried out using the
distributed model ‘InfoWorks-CS’. Serial rainfall monitoring data were used for inves-
tigation of water runoff behaviour from the highway. The inter-event time definition was
calculated from the runoff simulation and applied for better accurate runoff volume
determination. Under continuous rainfall condition, the quantity simulation showed a
good agreement with the measured hydrograph. However, in some cases where rainfall
was not continuous, overestimation of runoff volume and difference in peak height/
timing was found after rainfall halting. It was believed that the initial/depression loss on
road surface was recovered during the halting period of rainfall and caused the difference
in the runoff volume and peak height/timing difference in simulation. The consideration
of regenerated depression loss significantly improved runoff simulation results in the
on-and-off type rainfall events. This result showed that it was essential to estimate the
inter-event time definition to consider appropriateness of the initial or intermediate loss.

Keywords: Highway runoff; Initial/depression loss; Inter-event time definition;

Quantitative simulations

highway is an increasingly important environmental issue

1. Introduction
for urban communities.
In urban areas, road/highway runoff is considered as one of The interest in highway runoff water quantity and quality
the major water runoff sources for its extensive impervious has increased the use of computer models (SWMM,
area and man-made watercourses. During rainfall many InfoWorks, MOUSE etc.) for the estimation and manage-
particles deposited on the road surface will be dislodged, ment of runoff water quality as well as quantity. However,
and many of them are toxic micropollutants such as heavy it is still difficult reliably to model runoff quantity and
metals and polycyclic aromatic hydrocarbons (Hoffman quality, although several researches have been carried
et al. 1984, Sansalone and Buchberger 1997, Furumai et al. out (Tomanovic and Maksimovic 1996, Bartosova and
2001). These micropollutants are reported to have con- Novotny 1999, Hijioka et al. 2001). Many of the models
siderable impacts on aquatic community such as lakes use advanced and sophisticated algorithms to represent
and streams (Baekken 1994, Maltby et al. 1995). Thus the various runoff phenomena and complicated flow routing.
management of quantity and quality of runoff from road/ The unique feature of each rainfall event and dynamic

*Corresponding author. Email: [email protected]

Urban Water Journal

ISSN 1573-062X print/ISSN 1744-9006 online ª 2007 Taylor & Francis
http://www.tandf.co.uk/journals
DOI: 10.1080/15730620601145873
 54 R. K. Aryal et al.

behaviour of runoff water makes the model deeply rely on sedimentation and two retention ponds), and in the case
the accurate field monitoring data. Among the data, the that the runoff flow rate exceeds 500 l/s, the excess flow is
inter-event time definition (IETD) and the recovery of discharged from the outfall O2.
depression loss are very important along with the rainfall A long-term continuous runoff monitoring was con-
and runoff data in obtaining the accuracy in the runoff ducted from September to December 2000 at the study
modelling. The data are more sensitive to road/highway area. Recorded rainfall height was from 0.8 mm to 67 mm
runoff where the pollutants load are very high in less runoff during the monitoring period. The runoff flow rate was
volume compared with other non-point sources and the measured every minute at the outfall. During the monitor-
surface becomes dry more quickly owing to traffic activities. ing, the runoff flowrate did not exceed 500 l/s and the
However, both the IETD and the recovery of depression runoff flowrate was measured only at outfall O1. The water
Downloaded By: [Tokodai Kinokunya Co] At: 03:59 26 February 2007

loss have been paid less attention. Adams et al. (1986) sampler was programmed to initialize when flowrate
found IETD between 1 and 6 h appropriate for typical exceeded 8 l/s (0.34 mm/h) and then to stop when flow-
urban catchments in Canada. Hijioka et al. (2001) roughly rate decreased to 5 l/s based on the measured base flowrate
applied the IETD for 8 h when applied in road runoff in 0.7 l/s. Rainfall signal was recorded in every 10 min at a
Japan. Merz et al. (2006) selected IETD lasting 3 – 6 h rainfall monitoring station located near the highway
based on trial and error based on work in 337 catchments [figure 1(left)]. The road surface was made of asphalt.
in Austria. Chen and Adams (2006) applied IETD up to 6 h Surface type was considered all as impervious. The long-
for urban catchment in Toronto Bloor Street. term simulation was carried out with the distributed model
In the current study, we carried out highway runoff software InfoWorks-CS (version 4.5). The drainage net-
quantity simulation with a distributed model (InfoWorks work file was constructed based on the highway design data
CS) based on long-term field monitoring data. We cal- except for the ramp area (data not available). In the
culated the IETD and applied for the quantity runoff sections of the routes A1 and A4, the manholes were
simulation. Finally, we identified the critical role of inter- situated in 80 m intervals. Since the road in the ramp area
event time definition and the recovery of initial/depression was wider than other parts, each manhole was allocated in
loss for precise explanation of runoff quantity in simulation. 20 m intervals so that the subcatchment area was similar to
the main route sections. Figure 1(right) shows a simple
schematic diagram of the drainage network and runoff flow
2. Materials and method
direction.
The study area is a highway drainage system covering an
area of 8.4 ha in Winterthur, Switzerland [figure 1(left)].
The drainage area contained 67 manholes/nodes and 280 3. Results and discussion
gully pots. The study section of the highway was divided
3.1 Estimation of initial/depression loss
into two routes A1 and A4 with a length of 1.8 and 1.2 km
covering an area of 5.5 and 2.0 ha, respectively. The routes Initial/depression loss is one of the most sensitive para-
A1 and A4 are connected by a ramp with the length of meters in runoff simulations in frequently occurring light
0.6 km with the area of 0.9 ha. The drainage system has rainfall events. The initial loss plays an important role,
two outfalls O1 in the route A4 and O2 in the route A1. especially, in the road/highway runoff where the surface
Most of the runoff is discharged through the outfall O1 dryness occurs rapidly after the rainfall owing to high
connected to a treatment facility (oil floatation and traffic volume. Thus, a calibration by a single runoff event

Figure 1. Study area (left) and schematic diagram of drainage system (right).
 Inter-event time definition and recovery of initial/depression loss 55

may not be sufficient to verify the value of initial loss. quantity model as precise as possible before proceeding to
Figure 2 shows a relationship between the rainfall height the quality model that we precede in the next phase. In the
and the runoff height calculated from 25 events at the study runoff quantity model, initial/depression loss and base
area where each event was separated from another by flowrate should be given. Base flowrate was determined as
applying enter-event time definition 8. From the intercept 0.7 l/s from the field monitoring data during dry weather
of the x-axis, it is seen that the runoff occurs once the period. Similarly, the initial loss 0.7 mm and the inter-event
rainfall exceeds 0.7 mm. So, we considered 0.7 mm as the 8 h was applied for the quantitative simulations for number
initial loss for the runoff. of events.
Quantity runoff simulations for 18 rainfall events carried
out with 1 min step are shown in table 1. The degree of
3.2 Runoff quantity simulation
Downloaded By: [Tokodai Kinokunya Co] At: 03:59 26 February 2007

agreement between simulation and observation was

3.2.1 Calculation of inter-event time definition loss. judged in terms of total runoff volume and peak height/
Accurate knowledge of the quantity of runoff is required timing. For the total runoff volume, 15% of difference
to assess the impact of the runoff on the environment to was used as a threshold for primary categorization of
develop appropriate technologies such as the storm water the simulation results, while 5 min was set for peak
drainage type and storage tank. Thus the number of rainfall height/timing. Table 1 summarizes the rainfall runoff
events and runoff volume at a time is important. To cal- characteristics and simulation results of eighteen rainfall
culate the number of events in a year we should have clear events.
IETD to discrete the continuous rainfall into each event. Among the 18 simulations, eight of them represented
To determine the IETD we carried out time-series runoff similar runoff volume and peak height/timing to the meas-
simulations with initial loss 0.7 mm and no recovery of the urement. Four examples with good agreement are shown in
depression loss during halting period. The difference be- figure 4. The upper bar diagrams in the figures shows the
tween measured and simulated runoff volumes in 13 on- rainfall intensity in every 10 min interval. These events had
and-off rainfalls with different halting periods was plotted weak rainfall intensity (maximum rainfall intensity less
(figure 3). The x-axis and y-axis in the figure show the than 5 mm/h). We categorized the eight events as the first
rainfall-halting period and the difference in runoff volume group as shown in table 1.
between simulations and measurement, respectively. We Among the remaining ten simulations, five of them
found that up to 8 h (480 mins) the difference increases (2 September, 8 October, 26 October, 24 November and
with increasing halting period and then becomes almost 26 November) gave similar hydrographs (pattern) and
constant. This implies that the depression loss recovers flow peaks with the measurement at the beginning of events.
linearly until 8 h. The initial loss matched the stabilized However, the simulated peak height was higher than the
difference in runoff volume (60 m3) obtained from the measured one at the later stage. Figure 5 (left top and
rainfall runoff relationship (figure 2). bottom) shows examples of events on 2 September and
26 October. In these events, we found discontinuity of
rainfall with longer than 3 h (table 1). We believed that the
3.3 Runoff quantity simulation
halting of rainfall in the events caused the regeneration/
Owing to the importance of flow dynamics as the dominant recovery of depression loss owing to evaporation and
mechanism for transporting pollutants, we assure runoff splashing of road water to the banks by heavy traffic

Figure 2. Rainfall-runoff relationship during the mon- Figure 3. Determination of inter-event time definition
itoring period. (IETD).
 56 R. K. Aryal et al.

Table 1. Summary of runoff simulation.

Characteristics Measured versus simulated hydrographs

Difference
in outflow
Rain RV MRI PRV ADWP T RH Measured Simulated Peak
event (mm) (mm/h) (mm) (h) (h) (43 h) outflow (m3) outflow (m3) (%) M H/T Group

2 Oct 6.1 4.2 1.6 8 3.0 N 419 454 8 G G 1 (all good)

5 Oct 6.0 4.2 6.1 72 39.5 N 485 449 77 G G
24 Oct 2.1 3.6 14.0 293 3.5 N 114 120 5 G G
31 Oct 10.2 3.0 5.2 100 11.3 N 765 816 7 G G
Downloaded By: [Tokodai Kinokunya Co] At: 03:59 26 February 2007

3 Nov 9.6 4.2 1.6 62 11.0 N 592 666 13 G G

4 Nov 2.2 4.2 1.2 8 8.7 N 195 204 5 G G
14 Nov 11.0 3.0 2.2 180 13.3 N 974 900 78 G G
8 Dec 5.2 3.0 1.5 49 18.0 N 466 411 713 G G
2 Sep 3.7 6.0 1.5 15 4.0 Y 401 524 30 B B 2 (all bad)
8 Oct 5.3 4.8 10.1 42 35.0 Y 357 411 15 B B
26 Oct 5.2 4.8 2.1 43 19.0 Y 337 431 28 B B
24 Nov 2.3 1.2 5.4 34 12.8 Y 226 166 727 B B
26 Nov 7.2 6.6 2.3 26 16.8 Y 772 601 722 B B
4 Sep 9.0 10.2 12.3 25 3.6 N 965 698 728 B B 3 (others)
14 Sep 5.3 25.8 12.9 183 7.2 N 686 406 741 B B
20 Sep 67.1 16.2 4.5 40 30.0 Y 9568 5697 759 B B
6 Oct 10.1 33.0 4.2 10 1.3 N 797 680 715 B B
11 Oct 14.0 6.0 3.7 22 21.0 N 1433 1184 717 B B

*Note: RV ¼ total rainfall height, MRI ¼ Maximum rainfall intensity in the event (measured at 10 min interval), PRV ¼ previous rainfall height,
ADWP ¼ Antecedent dry weather period, T ¼ rainfall duration, RH ¼ rainfall halting (43 h), Y ¼ Yes, N ¼ No, Peak H/T ¼ Peak height/timing,
M ¼ Matching, G ¼ Good matching (Outflow 5 +15%, peak H/T 5 5 mins), B ¼ Bad matching (Outflow  +15%, peak H/T 5 mins).

Figure 4. Hydrographs of runoff events on 5 October (left top), 24 October (right top) and 31 October (left bottom) and
14 November (right bottom).

activities. The ignorance of such a recovery of depression To rectify the simulation result we introduced the
loss might have caused to increase the simulated peak height recovery of depression loss to the model by descritizing
as well as runoff volume. the event into two periods (i) before halting period
 Inter-event time definition and recovery of initial/depression loss 57

and (ii) after halting period. We simulated each period top and bottom) shows the simulation after introduction
considering it as a separate event and introduced of the recovery of depression loss. The introduction
recovery/depression loss in each event. Figure 5 (right of recovered depression loss decreased the simulated
Downloaded By: [Tokodai Kinokunya Co] At: 03:59 26 February 2007

Figure 5. Hydrographs of runoff event on 2 September and 26 October before (left top and bottom) and after accounting
the recovery of depression loss (right top and bottom) respectively.

Figure 6. Hydrographs of rainfall events on 4 September (left top), 14 September (right top) and 20 September (bottom).
 58 R. K. Aryal et al.

peak height and difference in outflow became less depression loss to explain runoff volume during the
than 5%. rainfall halting. It is necessary to consider the recovery
This indicated that the accounting of recovered loss is of depression loss for better estimation of hydrograph
very important for runoff simulation with on-and-off type in on-and-off type rainfall events.
rainfalls in the highway drainage. We categorized these (5) The runoff simulation of strong rainfall events failed to
events into the second group. match the measured and occurred always less runoff
The remaining five simulation results (4 September, volume than the measured indicating the possibility
14 September, 20 September, 6 October and 11 October) of spot/localized rainfall and/or runoff encroachment
showed difference in peak height/timing. These events from the side of the highway.
characterized by strong rainfall intensity (having intensity
Downloaded By: [Tokodai Kinokunya Co] At: 03:59 26 February 2007

greater than 10 mm/h) and categorized into the third

Acknowledgements
group as shown in table 1. Figure 6 shows examples of
hydrographs in events on 4 September (left) 14 September We would like to thank Professor M. Boller, Mr H. Balmer
and 20 September (right). The simulated hydrograph in and Mr J. Eugster (Swiss Federal Institute of Environ-
the event on 4 September and 20 September was almost mental Science and Technology (EAWAG)) for their kind
similar to the measured one except for the peak height, cooperation.
while there were differences in peak height and timing for
the event on 14 September. We believed that there are two
possible reasons for bad matching: References
Adams, B.J., Fraser, H.G., Howard, C.D.D. and Hanafy, M.S., (1986).
(a) there might be localized or spot rainfall so that the Meteorological data analysis for drainage system design. J. Environ.
monitored rainfall would not be representative to Engng, ASCE, 1986, 112, 827 – 848.
whole drainage area; Baekken, T., Effects of highway pollutants on a small Norweigan lake. Sci.
(b) there might be contribution of runoff from the side of Total Environ., 1994, 146/147, 131 – 139.
Bartosova, A. and Novotny, V., Model of spring runoff quantity
highway.
and quality for urban watersheds. Water Sci. Technol., 1999, 39,
249 – 256.
Chen, J. and Adams, B.J., A derived probability distribution approach to
4. Conclusion
stormwater quality model. Adv. Water Res., 2006 (in press).
A long-term runoff monitoring was carried out at a high- Furumai, H., Aryal, R.K. and Nakajima, F., Profile analysis of polycyclic
aromatic hydrocarbons and heavy metals in size fractionated highway
way drainage system in Winterthur, Switzerland from
dust and runoff samples, 9th International Conference on Urban
September to December 2000. The serial rainfall runoff Drainage (9ICUD), Portland, 2001.
data was used to perform the runoff quantity simulation Hijioka, Y. Nakajima, F. and Furumai, H., Modified models of wash-off
with a distributed model ‘InfoWorks CS’. The following from roofs and roads for non-point pollution analysis during first flush
conclusions could be made from the simulation. phenomena. Urban Drainage Modelling: Proceedings of the Specialty
Symposium of the World Water and Environment Resources Congress,
edited by W.R. Brashear and C. Maksimovic, pp. 275 – 286, 2001
(1) Under continuous rainfall conditions, the quantity (American Society of Civil Engineers: Reston).
simulation succeeded in explaining dynamic change of Hoffman, E.J., Mills, G.L., Latimer, J.S. and Quinn, J.G., Urban runoff as
hydrographs in frequently occurring light rainfall source of polycyclic aromatic hydrocarbons to coastal waters. Environ.
events. Sci. Technol., 1984, 18, 580 – 587.
Maltby, L., Forrow, D.M., Boxall, A.B.A., Calow, P. and Betton, C.I., The
(2) The rainfall runoff relationship showed the initial loss
effect of motorway runoff on freshwater ecosystems: 2. identifying major
0.7 mm. The continuous rainfall runoff data and toxicants. Environ. Toxicol. Chem., 1995, 14, 1093 – 1111.
simulated results showed the inter-event time defini- Merz, R., Bloschl, G. and Parajka, J., Spatio-temporal variability of event
tion 8 h for rainfall event. runoff coefficients. J. Hydrology, 2006 (in press).
(3) The recovery of depression loss was found almost Sansalone, J.J. and Buchberger, S.G., Partitioning and first flush of
metals in urban roadway storm water. J. Environ. Engng, 1997, 123,
linear with time until complete recovery (8 h).
134 – 143.
(4) In on-and-off type rainfall overestimation of peak Tomanovic, A. and Maksimovic, C., Improved modelling of suspended
height was found with rainfall halting. This showed solids discharge form asphalt surface during storm event. Water Sci.
that there existed a critical process of recovery of Technol., 1996, 33, 363 – 369.

View publication stats