Chapter 2

Runoff Generation Mechanisms

Copyright 2003 David G Tarboton, Utah State University

CHAPTER 2: RUNOFF GENERATION MECHANISMS

Figure 2 depicts a cross section through a hillslope that exposes in
more detail the pathways infiltrated water may follow. Infiltrated
water may flow through the matrix of the soil in the inter-granular
pores and small structural voids. Infiltrated water may also flow
through larger voids referred to as macropores. Macropores include
pipes that are open passageways in the soil caused by decaying roots
and burrowing animals. Macropores also include larger structural
voids within the soil matrix that serve as preferential pathways for
subsurface flow. The permeability of the soil matrix may differ
between soil horizons and this may lead to the build up of a saturated
wedge above a soil horizon interface. Water in these saturated
wedges may flow laterally through the soil matrix, or enter
macropores and be carried rapidly to the stream as subsurface
stormflow in the form of interflow.

Figure 2. Pathways followed by subsurface runoff on hillslopes.

(From Kirkby, 1978)

Rainfall-Runoff Processes

Chapter 2: 1

See Online Resource

Overview of processes
involved in runoff
generation

Recent research in hillslope hydrology involving tracers, especially in

humid catchments has found that the dominant contributor to
stormflow in the stream is pre-event water (averaging 75% world
wide, Buttle, 1994). Pre-event water is water that was present in the
hillslope before the storm as identified by a distinct isotopic or
chemical composition. Another consensus emerging from recent
research is that interflow involving preferential flow through
macropores is a ubiquitous phenomenon in natural soils. Rapid
lateral flow through a network of macropores and the effusion of old
water into stream channels is the primary mechanism for runoff
generation in many humid regions where overland flow is rarely
observed. This mechanism has been linked to nonlinear threshold
type behavior in hillslope runoff response. Figure 3 shows how
runoff ratio, the fraction of precipitation that appears as runoff, is
dependent upon soil moisture content. Soil moisture content needs
to exceed a threshold before any significant runoff occurs. Figure 4
shows the relationship between depth to groundwater and runoff at
two different hillslope locations (Seibert et al., 2003) that also shows
threshold behavior, with runoff being more tightly related to depth to
groundwater near the stream than further up a hillslope.

Figure 3. Relationship between runoff ratio and soil moisture

content (Woods et al., 2001, Copyright, 2001, American

Geophysical Union, reproduced by permission of American

Geophysical Union).

Natural soils contain heterogeneities that lead to variability in the

infiltration process itself. Infiltrating water follows preferential
pathways and macropores and may result in increases in moisture
content at depth before saturation or similar increases in moisture
content higher in the soil profile. Figure 5a shows a photograph of a

Rainfall-Runoff Processes

Chapter 2: 2

See Online Resource

Animation of Preferential
Pathway Infiltration

soil where dye has been used to trace infiltration pathways in

experiments reported by Weiler and Naef (2003). Figure 5b shows
the dye intensity objectively classified from the photograph following
excavation of the plot following a dye sprinkling experiment. Figure
5c shows moisture content over time measured at a range of depths
using time domain reflectometry in these sprinkler experiments.
Runoff [l/(s km)]
0

100

200

300

400

0.5

(b) 103 m from stream

Runoff [l/(s km)]
0

100

200

300

400

0.0
1.0

1.5

Depth to groundwater [m]

Depth to groundwater [m]

0.0

0.5

(a) 14 m from stream,

1.0

1.5

Figure 4. Relation between runoff and depth to groundwater for

two different locations in the Svartberget catchment (Seibert et al.,

2003, Copyright, 2003, American Geophysical Union, reproduced
by permission of American Geophysical Union)

With this background on the pathways followed by infiltrated water

we can examine the mechanisms involved in the generation of runoff
(Figure 6). Each mechanism has a different response to rainfall or
snowmelt in the volume of runoff produced, the peak discharge rate,
and the timing of contributions to streamflow in the channel. The
relative importance of each process is affected by climate, geology,
topography, soil characteristics, vegetation and land use. The
dominant process may vary between large and small storms.

Rainfall-Runoff Processes

Chapter 2: 3

Figure 5. (a) Photograph of cross section through soil following

dye tracing experiment (Courtesy of Markus Weiler).

Figure 5. (b) Objectively classified dye intensity following

sprinkler experiment. (Courtesy of Markus Weiler)

Rainfall-Runoff Processes

Chapter 2: 4

Figure 5. (c) Moisture content change measured using time

domain reflectometry during sprinkler experiment. (Courtesy of

Markus Weiler)

In Figure 6a the infiltration excess overland flow mechanism is

illustrated. There is a maximum limiting rate at which a soil in a
given condition can absorb surface water input. This was referred to
by Robert E. Horton (1933), one of the founding fathers of
quantitative hydrology, as the infiltration capacity of the soil, and hence
this mechanism is also called Horton overland flow. Infiltration
capacity is also referred to as infiltrability. When surface water input
exceeds infiltration capacity the excess water accumulates on the soil
surface and fills small depressions. Water in depression storage does
not directly contribute to overland flow runoff; it either evaporates or
infiltrates later. With continued surface water input, the depression
storage capacity is filled, and water spills over to run down slope as
an irregular sheet or to converge into rivulets of overland flow. The
amount of water stored on the hillside in the process of flowing
down slope is called surface detention. The transition from depression
storage to surface detention and overland flow is not sharp, because
some depressions may fill and contribute to overland flow before
others. Figure 7 illustrates the response, in terms of runoff from a
hillside plot due to rainfall rate exceeding infiltration capacity with the
filling of depression storage and increase in, and draining of, water in

Rainfall-Runoff Processes

Chapter 2: 5

See Online Resource

Animations of Infiltration
Excess Runoff
Generation

surface detention during a storm. Note, in Figure 7, that infiltration

capacity declines during the storm, due to the pores being filled with
water reducing the capillary forces drawing water into pores.
Due to spatial variability of the soil properties affecting infiltration
capacity and due to spatial variability of surface water inputs,
infiltration excess runoff does not necessarily occur over a whole
drainage basin during a storm or surface water input event. Betson
(1964) pointed out that the area contributing to infiltration excess
runoff may only be a small portion of the watershed. This idea has
become known as the partial-area concept of infiltration excess
overland flow and is illustrated in Figure 6b.
Infiltration excess overland flow occurs anywhere that surface water
input exceeds the infiltration capacity of the surface. This occurs
most frequently in areas devoid of vegetation or possessing only a
thin cover. Semi-arid rangelands and cultivated fields in regions with
high rainfall intensity are places where this process can be observed.
It can also be seen where the soil has been compacted or topsoil
removed. Infiltration excess overland flow is particularly obvious on
paved urban areas.
In most humid regions infiltration capacities are high because
vegetation protects the soil from rain-packing and dispersal, and
because the supply of humus and the activity of micro fauna create
an open soil structure. Under such conditions surface water input
intensities generally do not exceed infiltration capacities and
infiltration excess runoff is rare. Overland flow can occur due to
surface water input on areas that are already saturated. This is
referred to as saturation excess overland flow, illustrated in Figure 6c.
Saturation excess overland flow occurs in locations where infiltrating
water completely saturates the soil profile until there is no space for
any further water to infiltrate. The complete saturation of a soil
profile resulting in the water table rising to the surface is referred to
as saturation from below. Once saturation from below occurs at a
location all further surface water input at that location becomes
overland flow runoff.

Rainfall-Runoff Processes

Chapter 2: 6

See Online Resource

Animation of Saturation
Excess Runoff
Generation

(a) Infiltration excess overland flow

(also called Horton overland flow)

qo

f
(b) Partial area infiltration excess overland flow
P
Fraction of area contributing to
overland flow
P

qo

f
(c) Saturation excess overland flow
Variable source area
P

qo

qr

qs
(d) Subsurface stormflow

P
qs

(e) Perched subsurface stormflow

qs

1
izon
r
o
H
2
izon
r
o
H
Impeding layer

Figure 6. Classification of runoff generation mechanisms

(following Beven, 2000)

Rainfall-Runoff Processes

Chapter 2: 7

f0
f1

Figure 7. Rainfall, runoff, infiltration and surface storage during a

natural rainstorm. The shaded areas under the rainfall graph

represent precipitation falling at a rate exceeding the infiltration

rate. The dark grey area represents rainfall that enters depression
storage, which is filled before runoff occurs. The light grey

shading represents rainfall that becomes overland flow. The

initial infiltration rate is f0, and f1 is the final constant rate of

infiltration approached in large storms (from Water in

Environmental Planning, Dunne and Leopold, 1978).

In humid areas streams are typically gaining streams (gaining water by

drainage of baseflow from the groundwater into the stream) with the
groundwater table near the surface coincident or close to the stream
water surface elevation. This means that the water table near streams
is close to the ground surface, especially in flat topography, making
these near stream areas in flat topography particularly susceptible to
saturation from below. The extent of the area subject to saturation
from below varies in time, both at seasonal and event time scales due
to fluctuations in the depth to the shallow water table. This
variability of the extent of surface saturation is referred to as the
variable source area concept (Hewlett and Hibbert, 1967) and is
illustrated in Figures 8 and 9.

Rainfall-Runoff Processes

Chapter 2: 8

Geometrical considerations dictate that near stream saturated zones

will be most extensive in locations with concave hillslope profiles and
wide flat valleys. However, saturated overland flow is not restricted
to near-stream areas. Saturation from below can also occur (1) where
subsurface flow lines converge in slope concavities (hillslope hollows)
and water arrives faster than it can be transmitted down slope as
subsurface flow; (2) at concave slope breaks where the hydraulic
gradient inducing subsurface flow from upslope is greater than that
inducing down slope transmission; (3) where soil layers conducting
subsurface flow are locally thin; and (4) where hydraulic conductivity
decreases abruptly or gradually with depth and percolating water
accumulates above the low-conductivity layers to form perched zones
of saturation that reach the surface.

Figure 8. Map of saturated areas showing expansion during a

single rainstorm. The solid black shows the saturated area at the
beginning of the rain; the lightly shaded area is saturated by the
end of the storm and is the area over which the water table had
risen to the ground surface (from Water in Environmental

Planning, Dunne and Leopold, 1978)

Rainfall-Runoff Processes

Chapter 2: 9

Return flow (qr in Figure 6c) is subsurface water that returns to the
surface to add to overland flow. Return flow also occurs at places
where the soil thins, for example rock outcrops and may manifest in
the form of springs.
In areas with high infiltration capacities, interflow, or subsurface
storm flow is usually the dominant contributor to streamflow,
especially on steeper terrain or more planar hillslopes where
saturation excess is less likely to occur. A number of processes are
involved in rapid subsurface stormflow. These include transmissivity
feedback, lateral flow at the soil bedrock interface and groundwater ridging.

Figure 9. Seasonal variation in pre-storm saturated area (from

Water in Environmental Planning, Dunne and Leopold, 1978)

Rainfall-Runoff Processes

Chapter 2: 10

See Online Resource

Animation of Subsurface
Stormflow

Transmissivity feedback (Weiler and McDonnell, 2003) is illustrated

in Figure 10 and occurs when water infiltrates rapidly along
preferential pathways and causes the groundwater to rise to the point
where highly permeable soil layers or macropore networks become
activated and transmit water rapidly downslope. Much of the water
that drains from the soil matrix into the macropore network is preevent water. This mechanism results in a nonlinear threshold like
response as illustrated in Figures 3 and 4.

Figure 10. Schematic illustration of macropore network being

activated due to rise in groundwater resulting in rapid lateral flow.

Lateral flow at the soil bedrock interface (Weiler and McDonnell,

2003) illustrated in Figure 11, occurs in steep terrain with relatively
thin soil cover and low permeability bedrock, where water moves to
depth rapidly along preferential infiltration pathways and perches at
the soil-bedrock interface. Since moisture content near the bedrock
interface is often close to saturated, the addition of only a small
amount of new water (rainfall or snowmelt) is required to produce
saturation at the soil-bedrock or soil-impeding layer interface. Rapid
lateral flow occurs at the permeability interface through the transient
saturated zone. Once rainfall inputs cease, there is a rapid dissipation
of positive pore water pressures and the system reverts back to a slow
drainage of matrix flow.

Rainfall-Runoff Processes

Chapter 2: 11

See Online Resource

Animation of Perched
layer stormflow

So
il

Low permeable
bedrock

Figure 11. Rapid lateral flow at soil bedrock interface.

The processes involved in the generation of subsurface stormflow by

groundwater ridging are illustrated in Figure 12. An idealized cross
section of a valley with a straight hillslope is shown. In a simplified
situation with uniform soils the water table has an approximately
parabolic form, and soil moisture content decreases with increasing
height above the water table. The shaded areas represent graphs of
soil moisture at the base, middle and near the top of the hillslope (a)
before the onset of rainfall; (b) as an initial response to rainfall; and
(c) after continuing rainfall. Because (in a) before the onset of water
input the water table slopes gently towards the channel there will be a
slow flow of groundwater to maintain the baseflow of the stream.
With the onset of surface water input, water that infiltrates near the
base of the hillslope will quickly reach the water table and cause the
water table near the stream to rise, early in a storm. Further upslope
the soil is dryer and distance to the water table greater. It therefore
takes longer for infiltrating water to reach the water table and where
the water table is deep all the infiltrating water may go into storage in
the unsaturated zone and not reach the water table for many days
after the storm. The initial response to water input is therefore as
depicted in Figure 12b, where the water table has risen near the
stream but remained unchanged further upslope. The rising water
table near the stream causes an increase in the hydraulic gradient
between the groundwater and stream, and increased subsurface flow
into the stream results. This is subsurface stormflow, and is
frequently seen to be groundwater that has been displaced by the
infiltrating water, and is thus old or pre-storm water bearing the

Rainfall-Runoff Processes

Chapter 2: 12

chemical and isotopic signature of water in the hillslope prior to the

storm, which may be different from the chemical and isotopic
signature of overland flow from rainwater that has not infiltrated.
Measurement of chemical and isotopic signatures of stream water,
ground water and rain water is commonly used in hydrology as a way
of inferring hillslope flow pathways. After continuing rain (Figure
12c), the water table has risen to the surface over the lower part of
the hillslope and the saturated area is expanding uphill. Some water
emerges from this saturated area and runs down slope to the stream.
This is termed return flow. Direct precipitation onto the saturated
zone (DPS) forms saturation excess runoff as described above.
Rain

(a)

Water
table

Baseflow
0

Rain
(b)
Water
table
Baseflow + subsurface stormflow
0

Rain
(c)

Direct precipitation
on saturated zone

Water
table

Return flow
Baseflow + subsurface stormflow
0

Figure 12. Groundwater ridging subsurface stormflow processes

in an area of high infiltration rate. (redrawn following Water in
Environmental Planning, Dunne and Leopold, 1978)

Figure 12 illustrates a region just above the water table that was close
to saturation. This is known as the capillary fringe, and can play an
important role in runoff generation in certain situations. Capillary
forces due to the surface tension between water and soil particles act
to pull water into the soil matrix above the water table and maintain
the capillary fringe at moisture content very close to saturation. The

Rainfall-Runoff Processes

Chapter 2: 13

addition of a small amount of water can saturate this soil and cause
the water table to rise quite rapidly, resulting in subsurface stormflow,
surface saturation and saturation excess overland flow. The moisture
content in the capillary fringe can also be affected by the history of
wetting and drying of the soil, a phenomenon known as hysteresis.
When soil has been draining the moisture content tends to remain
above what it would be if it were filling at the same pressure. The
addition of a small amount of water can switch the soil from draining
to filling mode, enhancing the effect of the capillary fringe on the rise
of the water table and subsurface stormflow response. The capillary
fringe and hysteresis are discussed in more detail in Chapter 4.
The discussion thus far has focused on the main processes involved
in runoff generation on a hillslope. To complete the discussion on
runoff generation processes it is necessary to mention briefly some
other processes and factors involved. Interception of precipitation
by vegetation can play a significant role in reducing runoff, especially
in forested environments. Much intercepted water is eventually
evaporated back to the atmosphere (Figure 1). In some hydrologic
models, interception is sometimes modeled as an initial abstraction that
is subtracted from precipitation inputs before they are used in
infiltration or runoff calculations. In other hydrologic models
detailed representations of the interception, storage of water in the
canopy, throughfall or stem flow are used (e.g. Rutter et al., 1972).
Direct precipitation onto a stream or water body also contributes to
runoff as indicated in Figure 6. This is important in areas where the
water surface is extensive, as with lakes, reservoirs and floodplains
that are flooded, because in these situations runoff generation is not
delayed by the usual hillslope processes.
The freezing state of the soil, in regions where freezing occurs, also
plays a role in runoff generation. Infiltration capacity is reduced due
to frozen ground, depending upon the soil moisture content at the
time of freezing.
Fire results in water repellency by soils which reduces infiltration
capacity. One cause for water repellency is chemicals released during
a fire that are absorbed in the soil, and can make it water repellent for
months to years following a fire. The heat from fire also removes the
thin films of irreducable water adhered to soil particles by capillary
forces, disconnecting potential flow paths. Penetration of water into
macropores following a fire is limited due to this effect. High
temperatures in deserts have the same effect, adding to the tendency

Rainfall-Runoff Processes

Chapter 2: 14

for infiltration capacities to be lower in arid regions making them

more subject to infiltration excess runoff generation processes. This
water repellency due to fire has been implicated in many floods
following severe bush or forest fires.
Many of the runoff generation processes described depend on the
soil moisture status of the soil. This is referred to as the antecedent
conditions. Between storms (surface water input events), processes of
evaporation, transpiration, percolation and drainage serve to set up
the soil moisture antecedent conditions. Runoff generation
mechanisms and processes therefore depend not only on conditions
during storms, but conditions in advance of storms and a complete
understanding or representation of all the land surface hydrologic
processes is required to quantify the generation of runoff.
Recognition of this has led to the development of continuous
simulation models, such as the National Weather Service Sacramento
soil moisture accounting model that keeps continuous track of the
state of different soil moisture components for the modeling of
runoff. Detailed presentation of these models is beyond the scope of
this module, although key ideas are reviewed at the end of this model.
The discussion above has reviewed, in a conceptual way many of the
processes and mechanisms involved in runoff generation. These can
be quite complex, and when efforts are made to perform quantitative
calculations the devil is in the details. Each watershed or hillslope is
different, with different topography, soils and physical properties.
The challenge for hydrologic modelers is to balance practical
simplifications with justifiable model complexity and the knowledge
that many specific physical properties required for detailed hydrologic
modeling are physically unknowable. Our understanding of runoff
generation involves the movement of water through soil pores and
macropores. These flows follow the physical laws governing fluid
flow (Navier Stokes equations) but we can never know in sufficient
detail the flow geometry to make use of fluid flow theory and
ultimately have to resort to simplifications or parameterizations of
the runoff generation processes. In the remainder of this module the
astute reader will note discrepancies between the physical
understanding given above and mathematical descriptions used to
perform practical calculations. The mathematical descriptions,
although frequently complex, incorporate significant simplifications
relative to the field based conceptual understanding of how runoff
processes work. This gap between field based and model based
representations makes the subject of rainfall runoff processes a

Rainfall-Runoff Processes

Chapter 2: 15

fertile area for research to learn how to better model rainfall runoff
processes.
Figure 13 summarizes the main processes involved in runoff
generation, showing the interaction between infiltration excess,
saturation excess and groundwater flow pathways. Most rainfall
runoff models are organized around a representation similar to
Figure 13 involving partition of surface water input into infiltration
or overland flow, either due to infiltration excess or saturation excess.
Infiltrated water enters the soil regolith where it contributes to
interflow, percolates to deeper groundwater or is evaporated or
transpired back to the atmosphere. The quantity of water in the soil
affects the variable source area involved in the generation of
saturation overland flow. The deeper groundwater contributes to
baseflow and affects interflow through groundwater rise.
Evapotranspiration
Hortonian OF

Surface Water Input

Infiltration

capacity

Saturation

Variable
Infiltration source area
Soil regolith
Percolation
Deeper
groundwater
aquifer

Saturation OF

Return flow

Regolith subsurface flow

(interflow)

Aquifer subsurface flow

(baseflow)

Figure 13. Hydrological Pathways involved in different runoff

generation processes. Infiltration excess pathways are shown in

red. Saturation excess and subsurface stormflow pathways are

shown in blue. Groundwater and baseflow pathways in black and

Evapotranspiration is green. (Courtesy of Mike Kirkby)

Rainfall-Runoff Processes

Chapter 2: 16

See Online Resource

View the Chapter 2
Summary

Exercises
1. Given the topographic map from Logan Canyon below, indicate
the location where saturation excess overland flow is most likely
to be generated during rainfall (from labeled locations, A, B, C,
D, E): ____

E
D
C
B

2.

Infiltration capacity is:

A. The number of foreign spies that a country can tolerate
B. The rate of water input to a stream by subsurface flow
C. The fraction of watershed area contributing to overland flow
D. The maximum rate at which water can be absorbed into soil
E. The water holding capacity of surface depressions

3. Subsurface stormflow is likely to be larger in:

A. A steep narrow valley
B. A wide flat valley

Rainfall-Runoff Processes

Chapter 2: 17

See Online Resource

Do the Chapter 2 Quiz

References
Betson, R. P., (1964), "What Is Watershed Runoff?," Journal of
Geophysical Research, 68: 1541-1552.
Beven, K. J., (2000), Rainfall Runoff Modelling: The Primer, John
Wiley, Chichester.
Buttle, J. M., (1994), "Isotope Hydrograph Separations and Rapid
Delivery of Pre-Event Water from Drainage Basins," Progress in
Physical Geography, 18(1): 16-41.
Dunne, T. and L. B. Leopold, (1978), Water in Environmental
Planning, W H Freeman and Co, San Francisco, 818 p.
Hewlett, J. D. and J. R. Hibbert, (1967), "Factors Affecting the
Response of Small Watersheds to Precipitation in Humid Areas," in
Forest Hydrology, Edited by W. E. Sopper and H. W. Lull,
Pergamon Press, New York, p.275-291.
Kirkby, M. J., ed. (1978), Hillslope Hydrology.
Rutter, A. J., K. A. Kershaw, P. C. Robins and A. J. Morton, (1972),
"A Predictive Model of Rainfall Interception in Forests, 1. Derivation
of the Model from Observations in a Plantation of Corsican Pine,"
Agricultural Meteorology, 9: 367-384.
Seibert, J., K. Bishop, A. Rodhe and J. J. McDonnell, (2003),
"Groundwater Dynamics Along a Hillslope: A Test of the Steady
State Hypothesis," Water Resources Research, 39(1): 1014,
doi:10.1029/2002WR001404.
Weiler, M. and J. J. McDonnell, (2003), "Virtual Experiments: A New
Approach for Improving Process Conceptualization in Hillslope
Hydrology." Journal of Hydrology, in review.
Weiler, M. and F. Naef, (2003), "An Experimental Tracer Study of
the Role of Macropores in Infiltration in Grassland Soils,"
Hydrological Processes, 17: 477-493, DOI: 10.1002/hyp.1136.
Woods, R. A., R. B. Grayson, A. W. Western, M. J. Duncan, D. J.
Wilson, R. I. Young, R. P. Ibbitt, R. D. Henderson and T. A.
McMahon, (2001), "Experimental Design and Initial Results from the

Rainfall-Runoff Processes

Chapter 2: 18

Mahurangi River Variability Experiment: Marvex," in Observations

and Modelling of Land Surface Hydrological Processes, Edited by V.
Lakshmi, J. D. Albertson and J. Schaake, Water Resources
Monographs, American Geophysical Union, Washington, DC, p.201213.

Rainfall-Runoff Processes

Chapter 2: 19