[Next Section] [Previous Section] [Table of Contents]

Introduction

Several studies have found a correlation between increased urbanization and changes in peak flow rates of streams (Barringer and others, 1994; Booth, 1988; Brown, 1988; Changnon and others, 1996; Ferguson and Suckling, 1990; Hollis, 1975; Kibler and others, 1981; Proceedings of the ASCE, 1969; Sauer and others, 1983; Wong and Chen, 1993). The United States Geologic Survey (USGS) delineated a method of estimated flood discharges in ungaged stream basins (Sumioka and others, 1997), but warned users to use a different method (Sauer and others, 1983) for urbanized basins. These effects are significant for small peak flow events, but are less significant for large events (Hollis, 1975; Sauer and others, 1983).

However, increased peak flows are not a universal response to urbanization. Basin characteristics and development patterns play a role in basin response to urbanization. Peak flow rates may actually decrease in some circumstances (Laenen, 1980).

Many basins in the Puget Sound-Willamette Valley lowland area are being increasingly developed. Laenen (1983) found changes in stream behavior after urbanization in the Willamette Valley. Urbanization has continued, as have questions about its continued effect on peak streamflows. The Portland area regional government, Metro (1998), found that of the 1,000 homes built near floodplains in the Metro area between 1992 and 1995, almost 200 were flooded in major flooding of February 1996.

Booth (1988) found that 100-year flows nearly doubled when impervious surfaces covered more than 20 percent of the area in small (0.3 to 18 km²) basins in western Washington. Hollis (1975) found a similar increase in 100-year flows with impervious surfaces of more than 20 percent of a basin. In addition, the number of smaller events (1 to 5 years) increased several times. Laenen (1983) found that a change from rural to urban conditions in the Willamette Valley increased storm runoff volume by 200% and peak flows by 300%.

To understand the reasons for these changes, a basic explanation of basin hydrology is needed. In its natural state, the hydrology of a river basin includes several pathways. When precipitation falls, interception and transpiration by vegetation return some water to the atmosphere. Water that reaches the surface infiltrates to the extent allowed by the conductivity and porosity of soil. Excess water ponds or runs across the surface of the ground, and either evaporates or congregates in stream channels. With additional ground and surface water in the basin, the flow of the river increases. Each of these elements is discussed below.

Dingman (1994) summarized canopy interception losses, and estimated that in Pacific Northwest Douglas-fir (Pseudotsuga menziesii) forests, from 24 – 35% of precipitation is lost to interception.

Factors influencing the infiltration rate include steepness of slope, roughness of surface, rate of precipitation, depth of surface ponding, saturated hydraulic conductivity at the surface, and water content of surface pores (Dingman, 1994). The most critical variables affecting infiltration are soil porosity and conductivity, and antecedent soil moisture (Lazaro, 1990). Porosity is a measure of the pore space in a volume of soil, and is determined primarily by the composition, size, and shape of minerals in a soil (Birkeland, 1984). For example, the porosity of rounded, fine-grained quartz sand (43%) is higher than that of glacial till which is also made primarily of sand (31%). Peat, which is primarily organic material, has a porosity of 92% (Bedient and Huber, 1992).

Hydraulic conductivity is the rate at which water moves through a soil. Porosity quantifies how many spaces there are, but conductivity measures how connected the spaces are. Conductivity is a rate (m/s) and is largely determined by grain size. Sandy soil has a hydraulic conductivity rate of 1.76 x10^-2 cm/s, while a soil made up of much smaller clay grains has a rate of 1.28 x 10^-4 cm/s (Dingman, 1994).

In the vadose zone, the unsaturated area between the soil surface and the groundwater table, water molecules are suspended among pores of soil by their attraction to solid surfaces and each other. The outer edge of the water film is most loosely held because of lower surface tension, and migrates down under the force of gravity. If the water input (precipitation) is equal to or less than the saturated hydraulic conductivity of a soil, no surface ponding and thus no overland flow will occur (Green and Ampt, 1911).

During rain events, however, water input may exceed the infiltration capacity of the soil. In some cases, there may be ground features that are relatively impermeable (e. g. rock faces). For an interim period, precipitation may remain on the ground surface, stored in depressions. This water must either evaporate or infiltrate. Precipitation that is not held in surface depressions becomes overland flow and joins the stream channel.

There is a time lag between the initial precipitation event and the first rise of the stream. During this period, the primary activities are interception by vegetation, infiltration, and travel of overland flow to streams. As the vegetation canopy and soil reach saturation, more water input may cause overland flow to occur, a mechanism referred to as Hortonian overland flow. In many humid watersheds, overland flow may not occur due to soils with high infiltration capacities. Subsurface flow is then the primary mechanism by which streamflow is generated by the watershed. Subsurface flow during storms may emerge from the soil in depressions, resulting in saturation overland flow (Dunne and Black, 1970). When the water input ceases, the basin begins to return to equilibrium conditions: water held in the canopy evaporates, water continues to move through the upper levels of soil until the soil is again capable of infiltration from a new precipitation event (Dingman, 1994).

In an urbanized basin, several parts of this process are disrupted (Geiger and others, 1987). Urban areas may create higher rainfall events, increasing the water input into a basin (Changnon, 1980). Transpiration declines as trees and shrubs are cut down and replaced by structures or groundcover such as grasses and barkdust. Evaporation rates change. Impermeable surface area increases, as buildings, parking lots, and roads cover soil. The resulting gutters, storm sewers and drains route water input to stream channels more quickly than the infiltration mechanisms they replace. Artificial depression storage areas, like improperly constructed parking lots, may inadvertently be created. Even when soil is not covered, compaction during construction activities can decrease the soil’s infiltration ability by reducing porosity and conductivity. Development of urban areas requires stream channels that are relatively stable, so river beds are often straightened, deepened, lined or channeled underground within urban areas, all of which have the effect of reducing infiltration and speeding water downstream. These effects may create increased peak flows in downstream areas that are not yet urbanized. The time necessary for the water input to flow through the basin decreases (Bedient and Huber, 1992). In effect, a stream in a temperate climate begins to react more like streams of an arid climate prone to flash floods. The tendency for streams to rise quickly during storms is referred to as the increased flashiness of a stream (Figure 1).

Figure 1. Unit hydrograph of Brays Bayou as land use changed. The shape of the hydrograph changed from a broad plateau in 1941 to a high, narrow peak after 50 years of development, showing a more flashy basin response (From Bedient and Huber, 1992).

As urbanization modifies basin characteristics, water flow shifts from a subsurface-dominated to a surface-dominated regime. As peak flows increase in size and number, increased flow velocities during runoff lead to increased bank erosion and scouring of river beds. Because water moves through the basin more quickly, there is less infiltration, reducing groundwater. Baseflows in low flow periods can decrease because of the depleted groundwater (Geiger and others, 1987).

The greatest effect on streamflow from urbanization is the disruption of infiltration. Infiltration rates can be calculated at the site-specific level by using soil characteristics and precipitation rates. At a basinwide level, however, the calculations become more difficult. Dingman (1994) describes Manley’s simple method of estimating basin runoff rates as:

q=w²/(2K_hsat)

where q is the runoff rate in m/s, w is the rainfall rate in m/s, and K_hsat is the maximum saturated hydraulic conductivity in m/s. As impervious surface area increases in extent, K_hsat decreases, w stays constant, so q must increase (Figure 1).

Stream basins respond to every precipitation event through the processes of interception, infiltration, evaporation, transpiration, and runoff. Changes in basin response can be observed by comparing streamflow variables to precipitation. Two variables can be used as indications of basin response: instantaneous peak flow and partial event volume.

Instantaneous peak flow (Q_p) is the highest flow (m³/s) of a peak event. Partial event volume (Q_v), as defined in this study, is not truly a volume, but is a summation of peak flows (m³/s) over the predefined duration of a peak event. An antecedent precipitation index (API) is a summation of precipitation (mm) over a predetermined number of days before the peak event (Figure 2). Fedora and Beschta (1989) found an API to be a useful tool for predicting storm runoff in the Oregon Coast Range.

Figure 2. An example of peak flow (Qp), partial event volume (Qv) and antecedent precipitation indices (APIs).

Changes in Q_p, Q_v, or APIs, or in relationships between these variables, reflect changes somewhere in the hydrologic system (Table 1). For example, increases in Q_p indicate the basin is producing higher peak flows, and increases in Q_v indicate the basin is producing a higher total volume of runoff during the event. Changes in Q_p or Q_v only reflect a basin effect but the cause(s) may be from several sources, including land use changes, geomorphic changes, and climate variation.

The effect of changes in precipitation can be eliminated as a potential cause if Q_p and Q_v are compared to APIs. If the relationship of peak flows or volume to antecedent precipitation (Q_p/API or Q_v/API) increases, it indicates that peak flows or volumes have increased after precipitation events of a similar size. This is the definition of a changed basin response.

Another indication of a changed response would be a change in the relationship between peaks and volumes (Q_p/Q_v). If this ratio increases, it suggests a quicker or flashier streamflow response to precipitation, because higher peaks (Q_p) were created but the total volume of runoff (Q_v) did not increase proportionately. An even stronger indication of an increased flashy basin response is the comparison of Q_p/Q_v to antecedent precipitation ((Q_p/Q_v)/API). An increase in this ratio explicitly shows an increased flashy response to similar amounts of precipitation.

In this study, four basins were examined to test for the effects of urbanization (Figure 3): the Tualatin River (1807 km² drainage basin) and Johnson Creek (68 km² drainage basin), both near Portland, Oregon; the Luckiamute River (614 km² drainage basin) near McMinnville, Oregon; and the Newaukum River (397 km² drainage basin) near Centralia, Washington.

The null hypotheses for this study were that from pre-urbanization to post-urbanization, at the 95% confidence level (a =0.05), there have been no changes in streamflow characteristics including:

· Peak flows (Q_p),

· Partial event volume (Q_v),

· Ratio of peak flow to partial volume (Q_p/Q_v), or

· Ratio of these values to an antecedent precipitation index (Q_p/API, Q_v/API, ((Q_p/Q_v)/API)).

Because of disruptions to the natural drainage system in urbanizing basins, it was expected that at least one of the null hypotheses would be rejected.

Table 1. Expected behavior with change in ratios of variables (Q_p=peak flow, Q_v=volume, API=antecedent precipitation index).

Figure 3. Site map of basins in study.

[Next Section] [Previous Section] [Table of Contents]