ETL 1110-2-540
30 Sep 96
(6) Immediately after data reception and filing, the data
warnings because errors caused low forecasts. Missing data
are checked for possible alarm conditions. If excessive rain-
must also be handled properly. A smoothly operating system
fall intensities or rapidly rising stream levels are detected, an
shouldn't be halted simply because data from a gage for a
alarm can alert the system operator.
particular time period are missing. Missing data should be
estimated from surrounding information and/or the operator
(7) A variety of displays can be used to present infor-
should be requested to intervene.
mation to users visually. Data reports, graphs, and maps are
commonly used. These displays can be shown on high
(1) Forecast models. The purpose of a forecast model is
resolution color monitors or sent to a printer to obtain a
to estimate future river flows and elevations based on
permanent copy.
observed or forecast amounts of rainfall. In flash flood
situations, certain portions of the forecast hydrograph are
(8) Text reports are used to summarize key data in
more important than others. Accurate forecasts of the rising
useful time increments such as 30 min, hourly, daily, etc.
limb, the time to hydrograph peak, and the magnitude of the
Graphs are used to examine trends. Maps are used to
peak are critical. These are the elements of model output that
examine the geographical variation of data. Each is bene-
have the most impact on the flood warning. The model
ficial in its own way to convey information to the user.
implemented in a flood warning system must consistently
perform well in these three areas.
(9) The following list (Federal Interagency Advisory
Committee 1985) summarizes the minimum features expected
(a) Before model selection, the very important element,
of software used in an automated flood warning system:
rainfall estimation, must be considered. The volume of water
under the rising limb of a flash flood hydrograph is primarily
Precipitation and stream gage data collection
surface runoff. Basins with short response times are often
!
characterized by low infiltration rates and steep slopes, which
Quality control of input data
efficiently generate runoff. Because these basins efficiently
!
generate runoff, especially during periods of high intensity
Historical databases
rainfall, the volume of runoff is very sensitive to the volume
!
of rainfall. This implies that the output of a flash flood
forecast model will also be very sensitive to the rainfall inputs.
!
Nemec (1984), for example, indicated that runoff simulation
Display of stream gage data
was 10 times more sensitive to a 5-percent change in
!
Visual and audible alarm based on excessive precip-
Sorooshian (1988) also suggests that, for intense storms, the
!
itation rate or rate of river rise
effect of rainfall variability on runoff behavior predominates
over that of the basin structure, so much so that the effects of
Hydrologic models (optional)
basin structure are difficult to separate.
!
Advisory forecast information (optional)
(b) Flash flood forecast sensitivity to rainfall inputs
!
serves to emphasize the importance of establishing a good
Link to NWS office
measurement system first. The phrase commonly heard in
!
the computer industry, "Garbage in, garbage out," is equally
f. Forecast Preparation System. The most common
applicable to flash flood forecasting. Good model perfor-
application programs in automated flood warning systems are
mance, no matter what model is used, cannot be expected
the runoff and river forecast programs. These programs
without a good measurement system. The implication for
utilize observed and, in some cases, forecast rainfall amounts
forecast system design is to invest in the measurement and
to compute runoff that will enter the stream system. Forecast
detection systems first, then consider hydrologic models.
programs usually update their results automatically, some-
(c) Many different hydrologic forecast models are in
times as often as every 5 min. When a forecast program
use. The most commonly used models in local flood warn-
determines that dangerously high streamflow may occur, an
ing systems fall into two categories: simple index-type
alarm can sound to alert the operator. Since forecast
models and conceptual rainfall-runoff models. Index models
preparation is automated, the system must be able to handle a
keep a running index that reflects current moisture condi-
variety of situations quickly. A robust data-error-handling
tions. The moisture index, a "time of year" index, current
capability must be incorporated to avoid generating warnings
rainfall, and rainfall duration is generally all that is needed to
based on erroneous data that is too high or generating no
estimate surface runoff with these models. Conceptual
4-10