FLOOD07101

• County: Statewide
• Basin: Statewide
• Title: A Numerical Modeling Study to Investigate the Assumptions Used in the Calculation of Probable Maximum Precipitation
• Date: 3/1/1999
• Prepared By: Water Resources Research
• Floodplain - Doc Type: Educational/Technical/Reference Information

,
I
I
,
I
I
I
J
i
j
~
j
I
1
I
I
I
,
i
1
,
1

i
I
I
j
I
I

ABBS: INVESTIGATION OF PROBABLE MAXIMUM PRECIPITATION ASSUMPTIONS

convective parameterization scheme on the finest grid to ini~
tiate convective development in locations dose to those ob.
served. For those cases the 7 km grid is too coarse to resolve
adequately the vertical motion scales most important to the
simulations. In these simulations the convective scheme is
phased out as the convection matures, following the method-
ology of Chen and Frank [1993]. Wang and Seaman [19971
discuss the use of a convective parameterization scheme for
simulations with a grid spacing of the order of 10 km and
report their findings for a number of case studies and several
convective parameterization schemes. They found that it is not
appropriate to use the explicit moisture scheme alone at these
model resolutions and that there were significant advantages to
using a convective parameterization scheme that directly rep-
resented moist, convective downdrafts.

3.2. East Coast Low: August 4-6

3.2.1 Validation. The model simulation commenced at
1200 UTC on August 4 and was run for the next 36 hours,
finishing at 0000 UTC August 6. The ECMWF mean-sea-Ievel
pressure analyses have been compared with the mean sea level
pre~sures predicted by the model for 0000 and 1200 UTC on
August 5 and 0000 UTC on August 6 (not shown). This com-
parison showed the rapid decrease in central pressure of the
low between 0000 and 1200 UTC and the close agreement
between the modeled and ECMWF pressures. At both of these
times the modeled pressures were within 2 hPa of the ECMWF
pressures, At 0000 UTC on August 6, there is less agreement
between the model and ECMWF analyses. with the model
beginning to increase the central pressure of the low and the
ECMWF analyses. indicating little change in the central pres-
sure of the low. It is believed that this erroneous 7 hPa pressure
rise is related to the prediction of the convection on the finest
mesh, as the rise does not occur in simulations with two levels
of grid nesling, Bolh the ECMWF analyses and the model
predictions show little horizontal movement of the low
throughout the final 24-hour period.
Figure 2 compares the observations of 0500 UTC, August 5,
with the model predictions corresponding to this time of the
wind field at the lowest model level (50 m) and the mean sea
level pressure. The features to note are the strong convergence
of the low-level winds at the leading edge (i.e., south of the
center) of the low and the weak winds that occur inland and
north of the low center. At this time. the modeled central
pressure of 1004 hPa compares favorably with the observed of
1002 hPa. The main difference between the model-predicted
and observed fields is that the model positions the low -60 km
further north and further offshore than the observed.
These discrepancies in the position of the low are reflected
in the precipitation predicted by the model. Figure 3 shows the
24-hour precipitation. ending at 2300 UTC on August 5. for
the observations and the model prediction. Both show a band
of heavy precipitation, of -300 mm, extending inland to the
Blue Mountains. The major difference between the two fields
is that the model predicts the band of heavy rainfall to occur
farther north than is observed, In addition, the northern
boundary of the rainfall band occurs farther north in the model
prediction. The orographic effect of both the Blue Mountains
nntJ the IIIawurru cscarpmcnt Is obvious in the local maxima in
the precipitation over both of these regions. As in the obser.
vations {Bureau of Meteor%gy, 1987J. the maximum rainfall
lies to the west of the lIIawarra escarpment and along the
southern escarpment of the Blue Mountains. The predicted

I
l
,
j

7SlJ

30

3S

I

14S 150 155

Figure 2. (a) Regional mean sea level pressure analysis
(hPa) at 05lHl UTC, August 5 [from Bureau of Meteorolo!.'Y.
1987] and (b) simulated regional mean sca level pressure anal-
ysis (hPa) at 0500 UTC, August 5, Results shown are for grid
2. Shading indicates terrain higher than 250 m with a contour
interval of 500 m.

rainfall amounts over both the Blue Mountains and the IIIa-
warra escarpment are very close to those observed. Inh:l1sily~
frequency-duration curves (not shown) have been calculated
for this simulation. These indicate that the model has been
able to capture the extreme intensity of the 8-hollr duratiun
precipitation.
When considering the quality of the model prediction, it
must be kept in mind that these predictions have bet:n pro-
duced at a horizontal resolution of -7 km from initial conJi.
tions that are available at a resolulion of - 250 km. Overall, the
model has predicted a realistic development of the cast coast
low and the predicted precipitation amounts and intensities ar~
close to the extreme values observed. ConlicqUl,mtly, WI.: h~lil,;v~
that this simulation provides an adequate test for many uf the
assumptions made in the estimation of PMP values.
3.2.2. Increased moisture simulations. The simulations
discussed in this section have been performed by initializing