Estimation of Time of Concentration For Three First Flush Highway Runoff Characterization Sites
Estimation of Time of Concentration For Three First Flush Highway Runoff Characterization Sites
Prepared for:
California Department of Transportation
Division of Environmental Analysis
1220 N Street
Sacramento, CA 95826
Prepared by:
2
Center for Environmental and Water Resources Engineering
Department of Civil and Environmental Engineering
One Shields Avenue, Engineering III
University of California
Davis, CA 95616
June 2006
1
TABLE OF CONTENTS
List of Figures................................................................................................................................... 3
Summary ................................................................................................................................... 4
SECTION 1 INTRODUCTION................................................................................................... 5
REFERENCES............................................................................................................................... 25
2
List of Tables
Table 7 Design Peak Flow Rates (qp) Calculated with Tc Values from Different Formulas…19
Table 8 Average MFF ratios for Site 1, Site 2 and Site 3 ........................................................23
3
List of Figures
Figure 2 Probability Plots for 24hr Rainfall Depth in Site1 with Different Partial
Duration Series............................................................................................................14
Figure 3 Probability Plots for 24hr Rainfall Depth in Site2 with Different Partial
Duration Series............................................................................................................15
Figure 4 Probability Plots for 24hr Rainfall Depth in Site3 with Different Partial
Duration Series............................................................................................................15
Figure 5 Unit Peak Flow Discharge with Respect to Tc for Different Ia/P Values
Figure 7 Correlations Between Time of Concentration and Mass First Flush Ratios
Figure 8 Correlations Between Time of Concentration and Mass First Flush Ratios
Figure 9 Correlations Between Time of Concentration and Mass First Flush Ratios
4
SUMMARY
Values of time of concentration (Tc) for three highway sites were estimated using different
formulas and the best design Tc was determined. Several Tc formulas available for impervious
watersheds were classified into two groups based on whether rainfall intensity is a variable in the
formula or not (Groups 1 and 2). Formulas in Group 1 are modified ASCE, Kirpich, FAA, and
SCS lag equation, returning a unique Tc value for each site. Group 1 formulas were evaluated
based on the calculated peak flow rates, and the best Tc estimation formula was determined for
each monitoring site. Five years’ rainfall data were used to prepare the partial duration series for
the frequency analysis and 2-year, 24-hour rainfall depths (P2) were obtained as a design rainfall
depth. A site survey was also conducted to obtain the longest flow length (L) and bed slope (S),
which are essential for the Tc calculation. The SCS lag formula was the best for the monitoring
sites, avoiding overestimation of design peak flow. Formulas in Group 2 are ASCE and Izzard
formulas, which include rainfall intensity as a variable in the formulas, as well as measured lag
time (distance between the centroids of hyetograph and hydrograph). Event-specific Tcs were
calculated by each formula in Group 2 using measured rainfall intensity, and the correlation
analysis for Tc and mass first flush (MFF) ratio were performed. Few relationships were
observed between Tc and MFF. However, when comparing the average MFF ratios and site-
specific Tc values from the three sites with different watershed areas, smaller watersheds tended
5
1. Introduction
conveyance, treatment, and disposal of runoff. Design of these facilities requires hydrological
and hydraulic information. Drainage systems such as drain-inlets, pipes, and channels are
designed to carry the maximum flow rate. The capacity of storage systems such as detention
basins or constructed wetlands can be determined based on the runoff volume. On the other hand,
concentration (Tc), which is used to determine peak flow rate, as well as flow patterns under
given rainfall characteristics. Tc is a site-specific parameter that depends on the rainfall and
Individual formulas have different domains of watersheds with different landuses or geometries
that the formulas were developed from. Therefore, it is important to choose an appropriate Tc
inappropriate formula is used (McCuen and Spiess, 1995; Cristina and Sansalone, 2003a).
Several different formulas are available to estimate Tc for highway landuses. Tc for the
sheet flow regime in a watershed is typically determined using equations of kinematic wave form
such as the ASCE formula and the SCS formula. Formulas from Izzard, Federal Aviation
Administration (FAA), and Soil Conservation Service (SCS) are also applicable for watersheds
In this study, Tc is calculated using different formulas that are considered applicable for
the highway runoff monitoring sites (Sites 1, 2, and 3) which are small impervious landuses
having relatively steep bed slopes. Using the Tc values, peak flow rates are calculated and
6
compared with real peak flow rates measured in the monitoring periods to evaluate and select the
best design value of Tc for each monitoring site. In addition, event-specific values of Tc are
calculated using measured rainfall intensity, and then compared with mass first flush ratios for
TSS and conductivity for each storm event in order to determine the relationships between water
2. Background
Tc is defined as the time required for a drop of water to travel from the most hydrologically
remote point in the catch basin to the point of collection. With uniform rainfall equally
contributed over a catch basin, outflow becomes equal to net input water after Tc, reaching
equilibrium (Viessman and Lewis, 2003). In a multi-flow segment catch basin, a representative
Numerous formulas to estimate Tc have been developed for different landuses and
of flood control systems such as drain pipes or inlets. The rational formula is the simplest
method for peak flow calculation. Assuming a uniform rainfall, peak discharge is calculated as
follows:
q p = CiA (1)
where qp = peak discharge (L3/T), C = runoff coefficient, i = design rainfall intensity (L/T), A =
catchment area (L2). Tc is considered a design rainfall duration to obtain i in equation (1), which
7
Table 1. Formulas for Time of Concentration (Tc)
1 L
SCS avg. vel. charts
(1975)
Tc = ∑
60 V
is the rainfall intensity corresponding to a design return period and duration in the intensity-
duration-frequency curve (I-D-F curve), which is developed using historical rainfall data. The
IDF curve for Southern California is available in Bulletin No. 195 published by Caltrans, DWR,
and FHWA in 1976. If iso-hyetal maps are available instead of I-D-F curves, SCS’s graphical
peak discharge method can be used for the peak flow calculation as follows:
q p = qu AmQFp (2)
where qp = peak discharge (cfs), qu = unit peak discharge (cfs/mi2/in), Am = drainage area (mi2),
8
Tc can also be used to predict flow changes for the design of drain inlets and pipes, and
for the capacity of treatment systems. The detail methods for these calculations are described in
Although flood control has been the main concern for hydraulic structure design, water
quality control has recently become an important issue in the stormwater runoff from highways,
which are known to generate significant amount of pollutants such as heavy metals, oil and
grease, and poly aromatic hydrocarbons (PAHs) (Roger et al., 1998; Furumai et al., 2002).
flush (FF) phenomenon, suggesting the emission of a greater fraction of pollutant mass or
concentration in the early part of the runoff volume (Ma et al., 2002; Sansalone and Cristina,
2004). This phenomenon enables compact best management practice (BMP) design with high
removal efficiency by treating only the earlier part of runoff. The first flush phenomenon is
catchment. Numerous efforts have been made to relate pollutant washoff behaviors with rainfall
intensity, flow rate, watershed area, or bottom slope, using statistical analyses of empirical
observations (Gupta and Saul., 1996; Deletic and Maksimovic, 1998; Cristina and Sansalone,
2003b). However, no clear relationship has been found. In addition, the performance of a BMP
depends on hydrologic and hydraulic conditions (Jacopin et al., 2001; Persson and Wittgren,
2003), which make it important to consider both water quantity and quality aspects in a BMP
design.
9
3. Methodology
As described in the previous section, a formula is basically used to calculate a unique value of Tc
(for design application), which is site-specific and determines the maximum capacity of
hydraulic structures. To calculate Tc for design application, design rainfall intensity or design
rainfall depths is required. However, event-specific Tc values can also be obtained using event
rainfall intensity instead of design rainfall intensity. The event-specific Tc values depend on the
hydrodynamic conditions of each storm event and therefore can be used for the correlation
In this study, Tc formulas were classified into two groups as shown in Table 2: one for
design Tc (Group 1) and the other for event-specific Tc (Group 2). Formulas in Group 1 are
modified ASCE, Kirpich, FAA, and SCS lag formulas and calculate Tcs for design applications.
These formulas do not require individual storm characteristics such as measured rainfall and
runoff. Single value of Tc for each site is calculated from each formula and used to calculate
design peak flow rate of the runoff. Group 2 includes ASCE and Izzard formulas and measured
lag time. ASCE and Izzard formulas commonly include rainfall intensity as a variable: as a result,
the iterative process is required to obtain Tc for design application. In this study, these formulas
were only used to calculate event-specific values of Tc using monitored average rainfall intensity
for individual storm events. The calculated Tc values and measured lag time were compared with
mass first flush (MFF) ratios for TSS and conductivity (e.g. MFF10, MFF20) for the monitored
storm events to investigate relationships between water quality and hydrological condition. TSS
and conductivity were selected as representative pollutants in particulate and dissolved forms.
10
Table 2. Tc Formulas in Two Groups
Formulas Purposes
Based on the ASCE’s kinematic wave analysis, the U.S. Federal Highway Administration
0.933L0.6 n0.6
Tc = (1)
i 0.4 S 0.3
where L = overland flow length (ft); n = Manning’s roughness coefficient (sec/ft1/3); i = the
rainfall intensity (in/hr); and S = the bed slope (ft/ft). In order to calculate Tc in equation (1), the
trial and error method is used by adjusting rainfall intensity until the calculated Tc matches the
storm duration corresponding to the applied rainfall intensity for the selected recurrence interval
in the I-D-F curve. To avoid the iterative calculation process, the Soil Conservation Service
11
where P2 = 2-year, 24-hour rainfall depth (in). To apply equation (2) in this study, design rainfall
depths for three sites were obtained by frequency analysis using five years’ monitoring rainfall
data.
To acquire site conditions and dimensions, the most recent construction drawings were collected
for the three first flush highway runoff characterization study sites. A site survey was also
conducted to verify the site slope, dimensions, and area for each site. Pertinent site dimensions
A stormwater monitoring program has been performed for six years from 1999 to 2005
and Table 4 summarizes the storm events used for the Tc calculation and correlation study. For
the frequency analysis, the hydrologic data gathered during the 1999-05 rainy seasons were used.
12
Table 4. Storm Event Summary (1999-2005 wet seasons)
13
3.3 Frequency Analysis of the Rainfall
Frequency analysis was conducted using partial duration series to obtain 2-year, 24-hour rainfall
depth (P2), which is required to calculate Tc in the modified ASCE equation (equation (2)). Three
different partial duration series were prepared by sorting out the three, four, and five largest
storms from each rainy season. P2 values obtained from the three different partial duration series
Figure 1 shows the frequency histogram of maximum 24-hr rainfall depth from the partial
duration series using five storms for each monitoring year for the monitoring sites. Event
distributions in Figure 1 were assumed to have lognormal distribution. Figures 2, 3, and 4 show
the graphical fits of lognormal distribution from different partial duration series (three, four, and
five largest storms selected for each monitoring year) for Site 1, 2, and 3, respectively. To obtain
the rainfall depth for the 2-year return period (P2), the following equation defining the return
1
= m[1 − F ] (3)
T
where T = return period, m = number of storms per year, and F = empirical estimate of frequency.
Using equation (3), F values for 2-year return period for 3-storm (m = 3), 4-storm (m = 4), and
5-storm (m = 5) partial duration series are calculated as 0.83, 0.88, and 0.90, respectively. P2
values corresponding to these frequency values on the regression line in Figures 2, 3, and 4 are
summarized in Table 5. As shown in Table 5, P2 values calculated from different series did not
significantly deviate from each other. Therefore, P2 from the 5-storm series was used in the Tc
calculation.
14
20 0.8 20 0.8
0.7 0.7
15 0.6 15 0.6
Count
Count
10 0.4 10 0.4
0.3 0.3
5 0.2 5 0.2
0.1 0.1
0 0.0 0 0.0
0 1 2 3 4 5 6 0 1 2 3 4 5 6 7
SITE1 SITE2
20 0.8
0.7
15 0.6
10 0.4
0.3
5 0.2
0.1
0 0.0
0 1 2 3 4 5 6
SITE3
Figure 1. Frequency Histograms of Maximum 24hr Rainfall Depth (in) for the Monitoring Sites.
S1_5 Storm
S1_4 Storm
0.5 S1_3 Storm
0
Log P
-0.5
-1
-1.5
.01 .1 1 5 10 20 30 50 70 80 90 95 99 99.9 99.99
F (%)
Figure 2. Probability Plots for 24hr Rainfall Depth in Site 1 with Different Partial Duration Series.
15
Partial Duration Series - Site 2
1
S2_5 Storm
S2_4 Storm
S2_3 Storm
0.5
Log P 0
-0.5
-1
-1.5
.01 .1 1 5 10 20 30 50 70 80 90 95 99 99.9 99.99
F (%)
Figure 3. Probability Plots for 24hr Rainfall Depth in Site 2 with Different Partial Duration Series.
S3_5 Storm
0.6 S3_4 Storm
S3_3 Storm
0.4
0.2
Log P
-0.2
-0.4
-0.6
.01 .1 1 5 10 20 30 50 70 80 90 95 99 99.9 99.99
F (%)
Figure 4. Probability Plots for 24hr Rainfall Depth in Site 3 with Different Partial Duration Series.
16
Table 5. P2 (2-year, 24-hour rainfall depth) for the Monitoring Sites from the Frequency
Analysis
Partial Duration Series
Monitoring Sites NOAA*
5-Storm 4-Storm 3-Storm
Table 6 shows values of Tc calculated by the modified ASCE, Kirpich, FAA, and SCS lag
formulas in Group 1. Each formula returns a single value of Tc for each site. The Kirpich
formula returns among the smallest Tc for all sites. This might be caused by the fact that the
Kirpich formula was developed for a steep slope area (3-10%); as a result, it underestimates Tc in
mildly sloped watersheds. In contrast, the SCS lag formula returns among the largest values of Tc.
This may be because this formula was originally developed for agricultural watersheds.
Formulas with the kinematic wave forms (i.e., ASCE, modified ASCE formulas) hold for the
sheet flow with steep bed slopes. Use of this formula is typically limited by a maximum flow
length of 300 ft to insure that the kinematic assumption is valid, although there is no documented
evidence that supports this criterion. McCuen and Spiess (1995) proposed the upper limit as
nL / S less than 100 (English units) from an empirical analysis using data from 59 watersheds.
Values of nL / S for Sites 1, 2, and 3 are calculated as 267, 81, and 41, respectively, using site
dimensions in Table 3 and 0.011 of n for smooth asphalt or concrete beds. This suggests that the
17
kinematic wave Tc formula (Equations (1) and (2)) might estimate well for Sites 2 and 3, and
might not for Site 1. The FAA formula returned slightly larger values of Tc than those by the
modified ASCE formula, with little difference when 0.95 of the runoff coefficient (C) is applied
Kirpich 7 3 2
Reduction factor (0.4) applied
for concrete or asphalt surface
Peak flow rates were calculated using different values of Tc as shown in Table 6, and evaluated
in terms of their capacity to accommodate actual peak flows generated during the storms events
in 1999-2005. SCS’s graphical peak discharge method was used. Design rainfall depth was 3.0 in
for all three sites, which was obtained from NOAA’s Atlas 2 maps (1973), showing an iso-hyetal
map of 2-year, 24-hour rainfall depth in west Los Angeles. A unit peak flow discharge graph for
type I rainfall area (Figure 5) was used to obtain the qu for the west Los Angeles area. An
18
• CN = 98 for all sites (pavement with larger than 95% imperviousness)
( P − 0.2 S ) 2
• Q (runoff) = = 2.77 in where, P = design rainfall (= 3.0 in)
( P + 0.8S )
o qu for site 1, site 2 and site 3 are 400, 500, 500 csm/in, respectively.
Figure 5. Unit peak flow discharge with respect to Tc for different Ia/P values (type I rainfall
distribution).
19
Table 7 summarizes peak flow rates and corresponding frequency (F) values for three
sites calculated using Tc values from different formulas through the procedure described above.
A value of F for a qp represent the probability of peak flow of a storm to be less than qp and were
obtained from the probability plots of six years’ monitoring events from 1999 to 2006 (figure 6).
Because all Tc formulas resulted in design qp values large enough to accommodate most of the
peak flows occurred for six years (F > 96%), A Tc formula resulting in less overestimated qp
should be used for cost-effective design. For example, qp obtained from SCS lag formula for site
2 (215 L/s) is among the smallest when compared to results from the other formulas, but still
much higher than the maximum qp occurred during the monitoring period. Therefore, for the
monitoring sites (site 1, site 2 and site 3), SCS lag equation, returning less overestimated peak
flow, was better than other formulas for the design purpose.
Table 7. Design peak flow rates (qp) calculated with Tc values from different formulas.
20
160
S1_peak flow
140 S2_peak flow
S3_peak flow
120
Mesured qp (L/sec)
100
80
60
40
20
0
.01 .1 1 5 10 2030 50 7080 90 95 99 99.9 99.99
F (%)
Tc is a function of watershed size and slope, which are also factors influencing mass emission
rate. Small watersheds with highly impervious landuse (e.g. highway, parking lot) are supposed
to have small Tc and usually have strong MFF in the pollutant emission (Ma et al., 2003;
Sansalone, 2004). Therefore, it is worth investigating relationships between Tc and MFF ratios.
formulas in group 2 (i.e. ASCE and Izzard formulas) using measured rainfall intensity for each
storm event and compared to the MFF ratios for TSS and conductivity, which were considered
most representative parameters for particulate and dissolved form of pollutants. Figure 7, 8 and 9
are the correlation charts showing the correlation among MFF10 and MFF20 for TSS and
conductivity, calculated Tcs and measured lag times for the storm events monitored in 2000-2003
21
for site 1, site2 and site 3, respectively. As can be seen, no clear relationship between Tc (or lag
time) and MFF ratios was observed. It is likely because there are other factors, independent of
Tc or lag time, but highly impacting on MFF such as antecedent dry days (ADD), antecedent
event rain, rainfall duration. Rainfall type can be also an influencing parameter for MFF, which
In addition, calculated Tcs are poorly related to the measured lag time. The poor
correlation between Tcs obtained from formulas and measured lag times may be due to the fact
that average rainfall intensity was used to calculate Tc, but temporal change of rainfall and flow
is a primary factor that determines lag time. Tcs from Izzard and ASCE formulas have a linear
Figure 7. Correlations between time of concentration and mass first flush ratios for TSS and
22
MFF20_COND MFF20_TSS MFF10_COND MFF10_TSS
ASCE
IZZARD
LAG_TIME
Figure 8. Correlations between time of concentration and mass first flush ratios for TSS and
Figure 9. Correlations between time of concentration and mass first flush ratios for TSS and
23
Table 8 shows the design Tc and the MFF ratios for TSS and conductivity from
monitoring sites. MFF10 and MFF20 larger than 1 were routinely observed in the storm events
with averages of 2.0, 2.0, 2.0 and 1.6 for MFF10 and MFF20 for TSS and conductivity,
respectively. Site 3, with the smallest Tc, has the largest MFF ratios among the three monitoring
sites. Site 1 has similar MFF ratio with site 2 with 2.5 times larger Tc. Although it was difficult
to obtain general relationship between Tc and MFF ratios, a watershed with small Tc (< 37 min)
usually have first flush, having higher MFF with smaller Tc.
Design Tc (min)* 37 11 6 -
* Obtained from SCS’s Lag formula
24
5. Conclusions
Available time of concentration (Tc) formulas were evaluated to determine design Tc values for
the three highway monitoring sites (site 7-201, 7-202 and 7-203) located in west Los Angeles. In
the frequency analysis to determine 2-year, 24-hour rainfall depths (P2) for the sites, partial
duration series using 3, 4, and 5 largest storms in the periods, 1999-2003, 2004-2005. Different
The Tc formulas were evaluated based on the capacity of resulting design peak flow. The
SCS’s lag formula provided the best Tc values with less overestimated peak flow. The values of
design Tc were 37, 11 and 6 minutes for site 7-201, 7-202 and 7-203, respectively. Design peak
flow rates were calculated using SCS’s graphical peak discharge method and larger than the
measured peak flows for 96% or more of the storm events occurred in six years from 1999 -
2006.
In addition, a correlation study for Tc and MFF was performed. Event-specific Tc values
were calculated using two Tc formulas (Izzard, ASCE) and compared with mass first flush ratios
(MFF10, MFF20) for TSS and conductivity for the storm events monitored in 1999 to 2005. No
clear relationship was found between Tc and MFF. However, when comparing average MFF
ratios and design values of Tc, site 3 with among the smallest Tc has higher first flush effect
25
References
Ahlfeld, D.P., and Minihane, M. (2004). “Storm flow from first-flush precipitation in stormwater
Cristina, C.M., and Sansalone, J.J. (2003). “Kinematic wave model of urban pavement rainfall-
Vidssman, W, and Lewis, G.L. (2003). Introduction to hydrology. 5th Edition, Prentice Hall, NJ.
Stedinger, J.R., Vogel, R.M., and Foufoula-Georgiou, E. (1992). “Frequency analysis of extreme
Roger, S., Montréjaud-Vignoles, M., Andral, M.C., Herremans, L. and Fortune, J.P. (1998)
Mineral, physical and chemical analysis of the solid matter carried by motorway runoff
Furumai, H., Balmer, H. and Boller, M. (2002) Dynamic behavior of suspended pollutants and
particle size distribution in highway runoff. Wat. Sci. Tech., 46(1), 413-418.
Jacopin, C., Lucas, E., Desbordes, M., and Bourgogne, P. (2001). “Optimisation of operational
management practices for the detention basins.” Wat. Sci. Tech., 44(2), 277-285.
Persson, J., and Wittgren, H.B. (2003). “How hydrological and hydraulic conditions affect
Sansalone, J.J. and Cristina, C.M. (2004). First flush concepts for suspended and dissolved solids
Ma, M., Khan, S., Li, S., Kim, L.-H., Ha, S., Kayhanian, M. and Stenstrom, M.K. (2002) First
flush phenomena for highways: how it can be meaningfully defined. Proceedings of 9th
26
McCuen, R.H. and Spiess, J.M. (1995). “Assessment of kinematic wave time of concentration.”
Gupta, K. and Saul, A.J. (1996) Specific relationship for the first flush load in combined sewer
Cristina, C.M. and Sansalone, J.J. (2003) "First Flush", Power law and particle separation
diagrams for urban storm-water suspended particulates. J. Environ. Engrg., ASCE, 129(4),
298-307.
Deletic, A.B. and Maksimovic, C.T. (1998) Evaluation of water quality factors in storm runoff
Ma, M., Khan, S., Li, S., Kim, L.-H., Ha, S., Kayhanian, M. and Stenstrom, M.K. (2002) First
flush phenomena for highways: how it can be meaningfully defined. Proceedings of 9th
Sansalone, J.J. and Cristina, C.M. (2004). First flush concepts for suspended and dissolved solids
U.S. Department of Agriculture, Soil Conservation Service, Engineering Division, 1985. SCS
U.S. Department of Agriculture, Soil Conservation Service, Engineering Division, 1986. Urban
27