34
ERDC TR-04-1
collect in arid regions, where rainfall data and data on physical watershed
parameters (e.g., soils) are spread over wide areas.
The lack of hydrological data makes empirical methods of flood estimation
in arid regions unsatisfactory, but physically based models that are calibrated in
the field using measurements of soil hydrological properties offer a viable alter-
native modeling strategy (El-Hames and Richards 1994). These limitations of
rainfallrunoff models can be overcome by using physically based input data that
can be integrated within a GIS framework to help with the management of the
large data sets required (Lange et al. 2000b). Parametric runoff models, while
lacking complexity, require large amounts of historical rainfallrunoff data for
calibration, data that are unavailable in ungauged watersheds. To overcome these
difficulties, a physically based model of an ephemeral watershed can be used to
generate synthetic calibration data (Truschel and Campana 1983). The processes
most important to account for in the rainfallrunoff modeling in arid regions are
infiltration, soil moisture storage, rainfall spatial distribution, evaporation, and
groundwater recharge (Niemczynowicz 1990). Drainage size, followed by rate of
infiltration and slope of land, are the most effective catchment characteristics
affecting the accuracy of some arid-region rainfallrunoff models (Nouh 1990).
Distributed rainfallrunoff models that account for spatial variations in rain-
fall amounts and soil types do a better job of modeling flash floods in semi-arid
regions than simple lumped methods (Michaud and Sorooshian 1994b). Other
studies highlight the importance of accurately defining temporal variations in
rainfall at a single point rather than establishing a dense network of rain gauges
to understand spatial distributions (Nouh 1990, Krajewski et al. 1991). Rainfall
runoff models developed specifically for arid regions can more accurately
account for transmission losses and other conditions characteristic of arid regions
(Sharma and Murthy 1998) and overcome the limited data sets from arid regions,
which inhibit the ability to calibrate such models.
A number of rainfallrunoff models, including HEC-1, have been accepted
by FEMA for use in the National Flood Insurance Program (Table 5). Despite the
limitations of modeling rainfallrunoff processes in arid regions, the accepted
FEMA models may serve as a starting point for identifying the models most
applicable for identifying the OHWM along ephemeral streams and for identify-
ing the limitations that must be overcome to create reliable rainfallrunoff
models for arid regions.
Previous Page
