1. Design Flood Estimation for Small Catchments in Southern Africa Using The Visual SCS-SA Software
Jeff Smithers and Roland Schulze
School of Bioresources Engineering and Environmental Hydrology
University of KwaZulu-Natal
Pietermartizburg
South Africa
Tel:

2. Introduction Your name Organisation
Background and expertise in design flood estimation
What you would like to learn this morning

3. Flood Drought Water in South Africa
We have either too much or too little!
Flood
Drought

4. Regional Scale Floods

7. Localised Floods Pietermaritzburg: 25 December 1999

8. Floods in Pietermaritzburg: 1987
Floods of 1987 (Pietermaritzburg)
RES2234

9. FLOOD HYDROGRAPHS FOR A SMALL CATCHMENT
Different Peak Discharges
Same Volumes
Discharge (m3.s-1)
Time (h)

10. FLOOD HYDROGRAPHS FOR A SMALL CATCHMENT
Same Peak Discharges
Different Volumes
Discharge (m3.s-1)
Time (h)

11. FLOOD HYDROGRAPHS FOR A SMALL CATCHMENT
Different Peak Discharges
Different Volumes
Discharge (m3.s-1)
Time (h)

12. FLOOD HYDROGRAPHS FOR A SMALL CATCHMENT
Significance of peak
capacities exceeded
flood damage (local)
Significance of volume
fills dams
transports sediments, nutrients etc
flood damage (regional, inundation)
Peak = f (volume)
Implication / Conclusion
need a model to simulate both stormflow volume and peak discharge
need to be able to simulate the entire hydrograph

13. Design Flood Estimation
What is a design flood?
Magnitude of flood which has acceptable risk associated with the failure of the hydraulic structures
Risk = probability of exceedance (Pe)
Return Period: T = 1/Pe
Not an observed event
What are design floods used for?
Design of hydraulic structures (e.g. waterways, culverts, bridges, dams etc)
How do we estimate design floods in South Africa?

14. Methods of Design Flood Determination in South Africa

15. Design Rainfall Event Based Models
SCS, Rational, Unit hydrograph
Widely used
Lump complex, heterogeneous catchment processes into a single process
Advantages
Simple to apply
Generally longer rainfall records at more sites, with better quality, than streamflow
Areal extrapolation of rainfall
Long flood series generally not available, often contain inconsistencies and are frequently non-homogeneous and non-stationary

16. Design Rainfall Event Based Models
Disadvantages
Uncertainties in inputs (e.g. storm duration, spatial & temporal distribution of design rainfall, model inputs)
Probability of rainfall taken into account, probabilistic nature of other parameters ignored
Antecedent soil moisture conditions
Assume that the exceedance frequency of the estimated flood = frequency of input rainfall
Design Rainfall Event Based Models
Widely used
Need to estimate design rainfall

17. WHY SCS-BASED DESIGN PROCEDURES?
There is frequent need for hydrological information recording
planning
design of water resources
management systems
For most small catchments the design hydrograph needs to be modelled / estimated because direct measurements are usually not available
SCS techniques were originally developed as a hydrological design tool on agricultural land uses
to generate safe limits in hydraulic design
to compare effectiveness of different agricultural/conservation systems
Reasons
equations are simple
related to physical properties of catchment (soils, land use, wetness)
provides uniform answers
uses daily rainfall input

18. WHY SCS-BASED DESIGN PROCEDURES?
Has become an accepted / established model on small catchments
procedure “used internationally several million times annually” (Hawkins, 1980)
recommended institutionally and accepted in court judgements
Tested / used widely in USA, Germany, France, mid-East, Australia, Africa
Now being used increasingly for other purposes through modification / adaptations, e.g.
daily water yield models
remote sensing inputs
environmental impact studies
urban areas
semi-arid areas
agricultural management systems
large catchments
South African adaptions
Regional differences in antecedent moisture conditions
Joint association of rainfall and runoff

19. Usage of surface runoff methods in SA (after Ward et al, 1989)
Question
Rational Method
(%)
Time-area (%)
SCS
Kinematic (%)
Other
Use of method 90 34 30 10 26
Reason for use
Easy to use 85 61 56 7 57
Not familiar with other methods 18 17 9
Method sufficiently accurate 55 50 38 46

20. Assessment of Methods Available for Small Catchments in SA (SRK, 1985)
Rational method
easy to use
but, peak discharge only
grossly overestimated peaks under all conditions
Time area method (e.g. Illudas model) & Kinematic Method (e.g. Witwat model)
neither performed consistently well
nor gave improved simulations considering increased model complexity
SCS based methods (esp. SA adaptations)
performed well enough on a number of land uses and catchment sizes to be recommended for design in South Africa

21. Background : Summary
There is frequent need for estimates of stormflow volumes (Q) and peak discharges (qp) from small catchments for making economic and safe design of hydraulic structures
Stormflow volumes and peak discharges are highly sensitive to a catchment’s “wetness” (i.e. antecedent soil moisture status, ASM) just prior to runoff producing rainfall events
The SCS technique has become a standard method for estimating Q and qp from small catchments (<30 km2)

22. HISTORY OF SCS METHODS IN SOUTHERN AFRICA
Concepts developed in USA in 1950s
Reich proposed its application in SA in 1962
Schulze & Arnold produced manual in 1979
Accepted / recommended by NTC, NPA, consultants
Considerable research effort at U of N, Pmb
Cousens (1976), Arnold (1980), Schulze (1982), Hope (1983), Schulze (1984), Schmidt & Schulze (1984), Dunsmore, Schulze & Schmidt (1986), Weddepohl (1988), Topping (1993), Chetty (2001)

23. HISTORY OF SCS METHODS IN SOUTHERN AFRICA
Water Research Commission / University of Natal contract 1984 – 1987
Update and revise SCS manual, integrating research findings
Research into joint association of rainfall and catchment moisture status to provide design runoff for different regions in SA
Production of manuals, technology transfer (700 sold)
Courses at 12 venues, 230 participants

24. HISTORY OF SCS METHODS IN SOUTHERN AFRICA
WRC and consultants requests led to development of PC version in 1992
400 sold; prescribed text; SA & IHE courses
Visual SCS-SA (2004)
Windows based, GUI
Regional scale invariance design rainfall estimation option added
SAIAE CPD 2004
SAIAE CPD 2005
Recent developments
An internationalised version, based on concepts developed in SA

25. The SCS Curve Number Model
SCS is the Soil Conservation Service of the USA Department of Agriculture
It works like this…
Rainfall occurs
Initial abstraction includes all losses to surface depressions, interception and initial infiltration
Then some water is infiltrated while some water is runoff

26. Source: http://www.fao.org/docrep/U8480E/U8480E3k.jpg

27. Runoff Two components SCS Stormflow (surface) Baseflow
Estimates stormflow only
Empirical equation with some physical basis

28. What factors influence stormflow depth ?
Rainfall
Depth
Intensity
Initial abstraction
Interception
Surface storage
Initial infiltration
Antecedent soil water
Soil properties
Infiltrability
Permeability
Storage capacity
Land cover
Type
Treatment, practice and condition

29. Depressional Storage &
SCS Curve Number Model
Initial Abstraction (accumulated losses before runoff commences)
Rate, Depth per Unit Time
Constant Intensity Rainfall
Infiltrated Water
Rainfall Excess
Runoff
Depressional Storage &
Interception
Constant Runoff
Evaporation
Time

30. STORMFLOW GENERATION WITH THE SCS : CONCEPTS

31. SCS Stormflow Volume Water balance Assume that ratio of
actual infiltration (F) to maximum retention (S)
is equal to
the ratio of runoff (Q) to potential maximum runoff (rainfall – initial abstraction)
Solve Equation 1 and 2 to estimate Q

32. The SCS Curve Number Model
Rainfall (P) measured or design amount
Initial abstraction (Ia) occurs from:
Surface depressions
Water intercepted by vegetation
Evaporation and infiltration
Potential maximum retention of soil (S)

33. The SCS Curve Number Model
Problem
P known
S & Ia unknown
Ia tends to be quite variable!
But after much experimentation:
Ia = 0.2 x S in the USA
Ia = 0.1 x S in SA
So substitute it back into the runoff equation:
But what about S ?

34. The SCS Curve Number Model
How can we estimate S?
where CN is a curve number according to the land use (from 0 to 100)
98 = Parking lot
39 = Grassed area on a very sandy soil
CN is an index of hydrological response
Use Table 5.1

37. Curve Numbers Index of catchment response How were CNs determined?
From measurements for given land cover a soil type
Plot of flood vs rainfall for annual maximum floods
Overlay of SCS stormflow equation for various values of CN
Median CN selected
CNs for “Wet” and “Dry” conditions determined and procedures for adjusting CNs for these conditions were developed

38. What Role do Soils Play?

39. What Role do Soils Play? Soil absorbs
Retains water
Releases water
Soil therefore a prime regulator of catchment response to rainfall
Evaluate soils from
agriculturalist
mechanical strength viewpoint
hydrological response

40. RES6511

41. RES6517

42. RES6519

43. RES6521

44. RES6523

45. Soil Categorisation in the Original USA SCS Model
There are 4 basic hydrological soil groups
Group A
Low runoff potential, high infiltration rates, sand, loamy sand and sandy loams
Group B
Moderate infiltration rates, loams, silt loams
Group C
Low infiltration rates, sandy clays
Group D
High runoff potential, very low infiltration rates, clay loams, clays, etc…

47. Hydrological Classification of Soils in SA
Wide spectrum of properties in South African soils
Four-fold grouping too course
Intermediate soil classification
A/B, B/C, C/D to give 7 groups

48. Soil Classification System in SA
Binomial System (Macvicar et al., 1977)
Soil form and series
Taxonomic System (SCWG, 1991)
Soil form, family and textural class

49. Hydrological Classification of Soils in SA
Classification procedure: Binomial System
Each soil placed in one of seven groups based according to the soils properties
Series graded up or down dependent on
Texture
Leaching
Water Table
Crusting
Classification procedure: Taxonomic System
Similar procedure

53. Taxonomic System (Table 5.2)

54. Binomial System (Table 5.3)

55. Sensitivity of Hydrological Response to Soil Properties
For a given catchment:
Area = 2 km2
Mean slope = 8%
Hydraulic length = 1500 m
Rainfall = 50 mm
Land use: veld cover : fair, i.e. plant cover %
Soil moisture status : initial
Clovelly Oatsdale (Cv16) : A/B
Stormflow depth : 1.73 mm Peak discharge : > 1 m3.s-1
Glenrosa Robmore (Gs18) : B/C
Stormflow depth : 9.22 mm Peak discharge : 4 m3.s-1
Estcourt Estcourt (Es36) : D
Stormflow depth : mm Peak discharge : 7 m3.s-1

56. Land Cover and Treatment
Land cover also makes a difference
Parking lots run off more than golf courses
Hydrologic condition makes a difference
Good or poor condition

61. Urban Stormflow

67. Sensitivity of Land Use on Hydrological Response
For a given catchment
Area = 2 km2 Rainfall = 100 mm
Response time (lag) = 0.5 h Intensity Distribution Type = 3
A/B : Griffin Farmhill, Veld in good hydrological condition
CN-II = 51
Stormflow volume = m3 Peak Discharge = 6.7 m3.s-1
A/B : Griffin Farmhill, Veld in poor hydrological condition
CN-II = 74
Stormflow volume = m3 Peak discharge = 19.2 m3.s-1
B : Clovelly Clydebank, Veld in good hydrological condition
CN-II = 61
Stormflow volume = 57000m3 Peak discharge = 11.4 m3.s-1
B : Clovelly Clydebank, Veld in poor hydrological condition
CN-II = 79
Stormflow volume = m3 Peak discharge = 22.8 m3.s-1

70. Adjustment of Initial Curve Numbers: Original Procedure
Stormflow is highly sensitive to a catchment's "wetness" (i.e. soil moisture status, SMS) just prior to the rainfall event

72. Adjustment of Initial Curve Numbers: Original Procedure
Stormflow is highly sensitive to a catchment's "wetness" (i.e. soil moisture status, SMS) just prior to the rainfall event
"Classical" categorisation of SMS
This is an oversimplification ….
ET considered only in gross terms
Drainage ignored
Discrete “jumps” in SMS vs CN
AMC – 5 days?
SMS by water budgeting needed
SMS class
Accumulated 5-day
Antecedent Rainfall
Dormant Season
Growing Season
SMS-I (CN-I)
< 12 mm
< 36mm
SMS-II (CN-II)
12-28 mm
36-53 mm
SMS-III (CN-III)
>28mm
>5 3mm

73. Adjustment of Initial Curve Numbers: Hawkins Procedure
ΔS = P – E –Q – D
Adjustment of CN-II therefore requires
CN-II
consideration of soil depth, soil texture, vegetation cover
regional climatic conditions

74. Adjustment of Curve Numbers to Account for Antecedent Soil Moisture Conditions in SCS-SA
Median Condition Method
Joint Association Method

75. Method 1: Median Condition Basic Premise
Final CN (CNf)needs to be determined from soil moisture budgeting considerations
Compute the soil moisture status expected (statistically) to occur most frequently at a location prior to a design event (50th percentile, median)
Use this SMS information in CNf calculations to determine design Q

76. Method 1: Median Condition Procedure
For a combination of input of
location (one of 712 hydrologically homogeneous zones in SA)
CN-II
soil depth category (one of 3)
soil texture category (one of 3)
vegetation category (one of 3)
Compute a change in soil moisture storage (ΔS)
by the ACRU model
from an initial soil moisture storage
for a 30-day antecedent period
for the 5 highest rainfall events of a year
for each year on record
The median condition of ΔS is computed

79. Method 1: Median Condition Computations
Use median ΔS to compute a final Curve Number, CNf
Use CNf to compute find potential maximum retention, Sf
Use Sf with design rainfall to compute final design stormflow

80. Method 2: Joint Association Method Basic Premise
Assumption that T-year return period rainfall produces T-year return period stormflow is invalid
2nd, 3rd, 4th or 5th ranked daily rainfall may produce highest annual Q, depending on antecedent SMS
Conclusion : "Assumption inherent in current flood design methods of simulating the T-year return period flood from the T-year return period rainfall does not provide the engineer with a sound basis for analysis in small catchments" (Dunsmore, Schulze, Schmidt, 1986)
Compute the highest daily Q per year, with a model, and use the series of simulated Q to determine design Q

81. Method 2: Joint Association Method Procedures & Computations
For a combination of input of
location (one of 712 homogeneous zones in SA)
CN-II
soil depth category (one of 3)
soil texture category (one of 3)
vegetation category (one of 3)
Compute a change in soil moisture storage (ΔS)
by the ACRU model
from an initial soil moisture storage
for a 30-day antecedent period
for the 5 highest rainfall events of a year
for each year on record

82. Method 2: Joint Association Method Procedures & Computations
For CN-II’s of 50, 60, 70, 80 and 90
Use ΔS to compute CNf for each of 27 land use/soil combinations
Calculate Qf for all combinations
Frequency analysis of Qf
50, 80, 90 and 95 perentiles
2, 5, 10 and 20 year return periods

87. Estimation of Daily Design Rainfall in South Africa
Option 1
Search database containing Adamson’s (1981) TR102 report which accesses a rainfall station information base for southern Africa
5 closest stations reported
Use select most appropriate
station

88. Estimation of Daily Design Rainfall in South Africa
Option 2
Design rainfalls up to 20 year return periods computed for the representative station chosen for each of 712 zones
Zone number determined from user input latitude and longitude
Option 3
User input design rainfall depths

89. Estimation of Daily Design Rainfall in South Africa
Option 4 (recommended)
Design rainfall estimated using a regional, scale invariance approach (Smithers and Schulze, 2003)
Methodology to estimate design rainfall at 1’ x 1’ lattitude/longitude grid in South Africa
durations 5 minutes to 7 days
2 to 200 year return periods
WRC reports
Visual SCS-SA
: 1 day design rainfall

93. What Factors Affect Peak Discharge?
Different Peak Discharges
Different Volumes
Discharge (m3.s-1)
Time (h)

95. Estimation of Peak Discharge Using SCS Procedures
Unit Hydrographs
The T-hour Unit Hydrograph (TUH) is defined as the surface runoff hydrograph resulting from a unit depth of effective rain falling uniformly in T hours over a catchment
Characteristic response from a catchment
Response is invariable
qp = f (Q)
SCS Procedures
Based on dimensionless Unit Hydrograph developed from large number of natural UHs
Shape of UH idealised to be triangular

96. SCS Triangular UH qp Tp Tr Tb

101. Time Distributions Of Design Rainfall Intensity
The timing and magnitude of peak discharge in relation to rainfall intensity
Small catchments
short catchment response time
short design storm duration critical (Why?)
high intensity storms are critical
Large catchments
long catchment response time
long design storm duration critical (Why?)
lower intensity storms
Regional design rainfall intensity
f (regional rainfall producing mechanisms)
f (regional synoptic conditions)
result in synthetic time distribution curves

102. Synthetic Time Distributions Of Rainfall Intensity In SA
One - day rainfall is distributed over time
Distribution assumed symmetrical over time
Element of conservatism built into procedures
Distribution based on PD : P24 h ratios
Four general types of time distribution curves identified for SA

104. Regionalisation of Temporal Distribution of Rainfall in SA

105. Synthetic Time Distributions Of Rainfall Intensity
Using SCS outside South Africa
Determine dominant design rainfall producing storms
Convective? : Type 3
General rains / frontal /
longer duration? : Type 2

106. Synthetic Time Distributions Of Rainfall Intensity In SA
Semi-stochastic rainfall disaggregation developed by Knoesen and Smithers (2005) not incorporated yet

115. Visual SCS-SA Visual SCS-SA IS
A computerised version of the 1988 SCS documentation for southern Africa
designed specifically for southern Africa,
but applicable (with limitations) universally
essentially a user manual
a "small" catchments design hydrograph technique
areas < 30km2
no ARF applied
where specific characteristics of
precipitation
land use
soils
physiography
dominate the hydrograph size and shape

116. Visual SCS-SA Visual SCS-SA IS NOT
A comprehensive flood estimation package for
multiple hydrographs
flow routing
An estimator of the PMF
A comprehensive theory document on the SCS techniques
A "large" catchments design hydrograph technique

117. Hands On Exercises