Chapter 2. Hydrology
33
Stream gauge data are too sparsely distributed in the Southwest to provide wide-
spread value in estimating peak discharges through statistical analysis.
A number of relatively simple techniques, such as the Rational Method and
the Curve Number Approach, are used by the Natural Resources Conservation
Service to establish crude runoff hydrographs. The Rational Method, originally
designed for watersheds less than 200 acres in area, assumes uniform rainfall
intensity over the whole watershed, a poor assumption in the Southwest, espe-
cially for larger watersheds, where thunderstorms are highly localized. The Curve
Number Approach attempts to determine the ratio of rainfall converted to runoff
by establishing a "curve number" that accounts for a number of controlling
variables including soil type and land use. Once a curve number is established for
a watershed, only the rainfall amount is needed to determine the depth of runoff
distributed over the entire watershed. Modeling programs developed by the
Natural Resources Conservation Service (e.g., TR-20) are available to convert
the distributed rainfall amount into a hydrograph using a dimensionless
hydrograph.
The most widely used rainfallrunoff model in the United States is HEC-1
(U.S. Army Corps 1998), which creates a synthetic flood hydrograph from user-
specified information about the precipitation event (e.g., duration, intensity),
existing conditions (e.g., antecedent moisture), and watershed characteristics
(e.g., drainage area, time of concentration, curve number). Even in arid regions,
accounting for deep groundwater and antecedent moisture conditions is critical in
modeling stream yields (Ye et al. 1997). A deptharea option in HEC-1 allows a
user to supply information on how precipitation depths vary within the drainage
area, a critical component for modeling flood hydrographs in arid areas where
precipitation totals are highly variable over small areas (Pilgrim et al. 1988).
However, even a relatively dense rain gauge network (one gauge per 20 km2)
may be insufficient to detect convective rainfall in semi-arid regions and estimate
its spatial coverage and depth (Michaud and Sorooshian 1994a). Transmission
losses typical of arid regions can be modeled in the flood routing component of
HEC-1 that accounts for channel and floodplain storage in various reaches along
the stream. General assumptions in the model regarding the shape of the hydro-
graph that apply well in temperate regions may not work as well in arid regions,
where a greater proportion of the flow is due to direct runoff (shorter time to
peak). As a result, the separate analysis of single events with gauge or radar
rainfall information is crucial for the accurate modeling of rainfallrunoff
characteristics of high-magnitude floods in arid catchments (Lange et al. 2000a).
While HEC-1 and other models are capable of simulating many of the conditions
present in arid regions, many of the input parameters needed are difficult to
Previous Page
