Sensitivity of orographic precipitation to changing ambient conditions:
An idealized modeling perspective
Brian A. Colle
*
Institute for Terrestrial and Planetary Atmospheres, Stony Brook University/SUNY
(includes mixed phase and graupel) explicit scheme and
MRF PBL (with no heat/moisture surface fluxes).Klemp
and Durran’s (1983) upper-radiative boundary condition
and a sponge layer were applied in order to prevent grav-
ity waves from being reflected off the model top. The 2-
D runs included rotational effects (f ~10
-4
s
-1
), therefore
terrain-normal flow can develop when a pressure gradi-
ent exists in the x-direction. Below are how the parame-
ters were varied:
U = 5, 10, 15, 20, 25, and 30 m s
-1
.
h
m
= 500, 1000, 1500, 2000, and 2500 m.
N
m
= 0.005 and 0.01 s
-1
.
L = 50 and 25 km.
FL = slightly below surface (sea-level temperature = 270
o
K, FLSFC), at 750 mb (FL750), and at 500 mb (FL500).
3. ROLE OF THE MOUNTAIN CIRCULATION
For the first set of experiments both L and FL were
fixed at 50 km and 750 mb, respectively. Figure 1 shows
the accumulated 6-12 h precipitation across the barrier
for a fixed ambient wind speed but slowly increasing
mountain height. As a result, the variations in Froude
number are determined by changes in barrier height. For
the simulations using U =10 m s
-1
and N
m
,=0.01 s
-1
(Fig.
1a), the precipitation distribution is broad and decreases
Figure 1. Accumulated precipitation (6-12 h) as a function of
terrain height (see inset box) for (a) 10 m s
-1
,N
m
= 0.01 s
-1
and
(b) 20 m s
-1
, N
m
= 0.01 s
-1
. (c) and (d) Same as (a) and (b)
except for N
m
= 0.005 s
-1
. The thick gray line is the mountain
location.
1. INTRODUCTION
Cool season precipitation forecasting in mountain-
ous terrain is challenging since the distribution and
amount of precipitation is controlled by a number of
dynamical and microphysical processes. It is commonly
knownthatmoistflow ascending a mountain barrier (i.e.,
upslope flow) will typically enhance the precipitation
along the windward slope. Both modeling (Sinclair
1994, among others) and observational (Pandey et al.
1999, among others) studies have shown that the magni-
tude of the upslope flow typically determines how much
precipitation will fall. However, there are other factors
that may determine the amount and distribution of oro-
graphic precipitation, such as the thermodynamic strati-
fication, moisture availability, wind profile above the
barrier, as well as hydrometeor advection and generation
rates. For example, recent studies have investigated the
role of flow blocking in enhancing the upstream oro-
graphic precipitation distribution as determined by the
Froude number, U/(h
m
N) (Sinclair 1994; Neiman et al.
2002), the importance of gravity waves in hydrometeor
production above narrow ridges (Bruintjes et al. 1994),
and the importance of microphysical timescales (Jiang
and Smith 2002).
There have been many idealized studies investigat-
ing the interaction of dry dynamics with topography at
all scales, butmuchlessattentionhasbeengiventomoist
flow dynamics and its influence on orographic precipita-
tion. In order to better understand the factors that effect
orographic precipitation, both 2-D and 3-D an idealized
modeling approaches are necessary. For many elongated
barriers, such as the California Sierras, Colorado Front
Range, and the France Pyrenees, 2-D numerical model-
ing has been shown to be an effective tool to understand
the terrain flow interaction; therefore, this approach was
utilized in many of our experiments. Some 3-D simula-
tions have also been completed for more complex terrain
such as the Wasatch Mountains of Utah and the Califor-
nia coastal range.
2. MODEL SETUP
The Penn State/NCAR mesoscale model (MM5v3)
was run in a 2-D idealized configuration using a 500 grid
point long domain with 4-km horizontal grid spacing, 39
sigma levels, and constant lateral boundary conditions.
The parameters defined during model initialization
include the barrier height (h
m
), mountain half width (L)
for a bell-shaped barrier, upstream flow velocity (U),
moist static stability (N
m
), and freezing level (FL). The
moisture initialization was specified to be nearly satu-
rated (99% relative humidity) throughout the domain,
therefore the initial static stability (N
m
) in this study is
defined relative to moist processes (Durran and Klemp
1982). The MM5 was integrated using the Reisner2
*
Corresponding author address: Dr. Brian A. Colle, Marine
Sciences Research Center, Stony Brook University, Stony
Brook, NY 11794-5000. email: bcolle@notes.cc.sunysb.edu.
6.3
50 100 150 200 250 300 350 400
0 5 10 15 20
50 100 150 200 250 300 350 400
0 5 10 15 20
50 100 150 200 250 300 350 400
0 5 10 15 20
50 100 150 200 250 300 350 400
0 5 10 15 20
50 100 150 200 250 300 350 400
0 5 10 15 20
2500 m
2000 m
1500 m
1000 m
500 m
50 100 150 200 250 300 350 400
0 5 10 15 20
Fr = 0 .5
Fr = 1.0
U = 10 m s
-1
, N
m
= 0.01 s
-1
km
mm
-->
h
m
50 100 150 200 250 300 350 400
0102030
50 100 150 200 250 300 350 400
0102030
50 100 150 200 250 300 350 400
0102030
50 100 150 200 250 300 350 400
0102030
50 100 150 200 250 300 350 400
0102030
2500 m
2000 m
1500 m
1000 m
500 m
50 100 150 200 250 300 350 400
0102030
Fr = 1.0
Fr = 2.0
U = 20 m s
-1
, N
m
= 0.01 s
-1
mm
50 100 150 200 250 300 350 400
0 5 15 25
50 100 150 200 250 300 350 400
0 5 15 25
50 100 150 200 250 300 350 400
0 5 15 25
50 100 150 200 250 300 350 400
0 5 15 25
50 100 150 200 250 300 350 400
0 5 15 25
2500 m
2000 m
1500 m
1000 m
500 m
50 100 150 200 250 300 350 400
0 5 15 25
Fr = 1.0
Fr = 2.0
U = 10 m s
-1
, N
m
= 0.005 s
-1
mm
50 100 150 200 250 300 350 400
0204060
50 100 150 200 250 300 350 400
0204060
50 100 150 200 250 300 350 400
0204060
50 100 150 200 250 300 350 400
0204060
50 100 150 200 250 300 350 400
0204060
2500 m
2000 m
1500 m
1000 m
500 m
50 100 150 200 250 300 350 400
0204060
Fr = 2.0
Fr = 4.0
U = 20 m s
-1
, N
m
= 0.005 s
-1
mm
km
gradually upstream of the barrier. As the barrier height is
increased from 500 to 1500 m the precipitation maxi-
mum shifts slightly upstream, while the maximum pre-
cipitation increase from h
m
= 2000 to 2500 m occurs
~150 km upstream of the crest because of flow blocking
(Fr < .5). Whentheflow is increasedto20ms
-1
(Fig. 1b),
more precipitation occurs over the upper windward slope
even for fairly low Froude numbers (Fr ≤ 1). Similarly,
for N=0.005 s
-1
and 10 m s
-1
(Fig. 1c,d), the precipitation
profiles for Fr ≤ 1 are peaked more over the upper wind-
ward slope than the Fr ≤ 1 runs for N
m
=0.01 s
-1
. Overall,
the shape of the cross barrier precipitation distribution is
more similar between runs of similar U or N
m
than Fr.
Figure 2a shows how the 6-12 h cross mountain pre-
cipitation profile for h
m
= 1500 m, N = 0.01 s
-1
, and L =
50 km varies as a function of ambient wind speed. The
thick arrow indicates where along the x-axis the precipi-
tation increases most rapidly for each 5 m s
-1
increase.
As the flow increases from 5 to 20 m s
-1
, there is down-
stream shift in precipitation towards the crest, with little
increase in precipitation more than 50 km upwind of the
crest. In contrast, for strong wind speeds (> 25 m s
-1
), the
precipitation enhancement shifts more upwind of the bar-
rier. Figure 3 presents the mountain circulation and pre-
cipitation species (rain, snow and graupel) for these
series of wind speed experiments in Fig. 2a. For
Figure 2. Accumulated precipitation (6-12 h) versus wind
speed (m s
-1
) for h
m
=1500 m for experiments of varying moun-
tain width (L), stability (N
m
), and freezing level (FL). The text
on panels a-e list the parameters used in each series of experi-
ments. The dashed line shows the mountain crest.
relatively weak flow (10 m s
-1
), the calculated and simu-
lated vertical wavelengths of the terrain-induced gravity
wave (λ
z
= 2πU/N
m
) are 6.3 and 7.5 km, respectively.
This wave results weak downward motion over the wind-
ward slope above 4 km and over the crest, which results
in a shallow orographic cloud over the windward side of
the barrier. When the wind speed is doubled (20 m s
-1
),
λ
z
doubles and the region of mountain-wave subsidence
over the crest rises to above 4 km, which allows for more
upward motion and precipitation above the upper wind-
ward slope and crest. For U = 30 m s
-1
(Fig. 3c), the large
vertical wavelength (~25 km) results in a deep area of
significant rising motion over the windward slope, thus
the precipitation area builds back upstream of the barrier
again. These systematic changes in mountain circulation
and cross mountain precipitation (Fig. 1) are similar for
other terrain heights since λ
z
is independent of h
m
. Over-
all, these results suggest the horizontal precipitation dis-
tribution for L=50 km is strongly effected by the vertical
structure of the mountain wave circulation.
Figure 3. Cross mountain profile (h
m
=1500 m; L = 50 km) of
potential temperature (thin solid), wind vectors, snow (gray),
graupel (dashed), and rain (solid) for U = (a) 10 m s
-1
, (b) 20 m
s
-1
, and (c) 30 m s
-1
.
50 100 150 200 250 300 350 400
0204060
50 100 150 200 250 300 350 400
0204060
50 100 150 200 250 300 350 400
0204060
50 100 150 200 250 300 350 400
0204060
50 100 150 200 250 300 350 400
0204060
50 100 150 200 250 300 350 400
0204060
30 m/s
25 m/s
20 m/s
15 m/s
10 m/s
5 m/s
N = 0.01 s
-1
, L = 50 km, hm = 1500 m
FL 750 mb
km
mm
50 100 150 200 250 300 350 400
0204060
50 100 150 200 250 300 350 400
0204060
50 100 150 200 250 300 350 400
0204060
50 100 150 200 250 300 350 400
0204060
50 100 150 200 250 300 350 400
0204060
50 100 150 200 250 300 350 400
0204060
30 m/s
25 m/s
20 m/s
15 m/s
10 m/s
5 m/s
N = 0.005 s
-1
, L = 50 km, hm = 1500 m
FL = 750 mb
mm
50 100 150 200 250 300 350 400
0204060
50 100 150 200 250 300 350 400
0204060
50 100 150 200 250 300 350 400
0204060
50 100 150 200 250 300 350 400
0204060
50 100 150 200 250 300 350 400
0204060
50 100 150 200 250 300 350 400
0204060
30 m/s
25 m/s
20 m/s
15 m/s
10 m/s
5 m/s
N = 0.01 s
-1
, L = 25 km, hm = 1500 m
FL= 750 mb
mm
50 100 150 200 250 300 350 40
0204060
50 100 150 200 250 300 350 40
0204060
50 100 150 200 250 300 350 40
0204060
50 100 150 200 250 300 350 40
0204060
50 100 150 200 250 300 350 40
0204060
(, , , )
50 100 150 200 250 300 350 40
0204060
30 m/s
25 m/s
20 m/s
15 m/s
10 m/s
5 m/s
N = 0.01 s
-1
, L = 50 km, hm = 1500 m
FL = 500 mb
mm
50 100 150 200 250 300 350 4
0
0 5 10 15 20
50 100 150 200 250 300 350 4
0
0 5 10 15 20
50 100 150 200 250 300 350 4
0
0 5 10 15 20
50 100 150 200 250 300 350 4
0
0 5 10 15 20
50 100 150 200 250 300 350 4
0
0 5 10 15 20
50 100 150 200 250 300 350 4
0
0 5 10 15 20
30 m/s
25 m/s
20 m/s
15 m/s
10 m/s
5 m/s
N = 0.01 s
-1
, L = 50 km, hm = 1500 m
FL = SFC
km
mm
(a) U = 10 m s
-1
(b) U = 20 m s
-1
(c) U = 30 m s
-1
km
Figure 4. Average precipitation 100 km upstream of the crest to
50 km in the lee for h
m
= 500 to 2500 m versus wind speed for
(a) N
m
=.01 s
-1
, L= 50 km and (b) N
m
=.005 s
-1
, L= 50 km.
Figure 4a presents the average precipitation for the
region 100 km upstream of the crest to 50 km in the lee
as a function of ambient wind speed for the L=50 km and
N
m
= 0.01 s
-1
. The precipitation amounts increase rap-
idly as U increases to 15 m s
-1
for many barrier heights,
but between 15-20 m s
-1
the increase slows as the moun-
tain wave begins building upstream. As the upward
motion with the vertical-propagating gravity wave deep-
ens over the windward slope for U> 20 m s
-1
(Fig. 3c),
the precipitation increases more rapidly again.
When N
m
is reduced to 0.005 s
-1
(Fig. 2b), the
upwind shift of precipitation increase occurs between 10
and 15 m s
-1
rather the 25 to 30 m s
-1
for N
m
=0.01 s
-1
.
This is consistent with λ
z
increasing more rapidly with
increasing wind speed for N
m
=0.005 s
-1
, resulting in a
more rapid buildup of precipitation upstream of the crest
compared to N
m
= 0.01 s
-1
. However, since the flow is
less stable for N
m
= 0.005 s
-1
, vertically propagating
gravity waves have less upstream tilt (φ) above the crest
for high wind speeds compared to N
m
= 0.01 s
-1
(N
m
cosφ = Uk
m
, where k
m
is the mountain wave number).
Thus, for less stable flow there is more upward motion
and precipitation over the crest at high windspeeds and a
downstreamshiftin precipitation towardsthe lee. For N
m
= 0.005 s
-1
the average precipitation over the barrier
increases nearly linearly with increasing wind speed for
moderate to high barriers (Fig. 4b). This is consistent
with recent observational studies under near moist neu-
tral conditions (Neiman et al. 2002).
4. ROLE OF MOUNTAIN WIDTH AND FREEZING
LEVEL
The mountain half-width (L) was decreased from 50
to 25 km to address the effects of mountain steepness for
N
m
= 0.01 s
-1
and FL = 750 mb. A steeper barrier favors
a more narrow precipitation distribution (Fig. 2c), with
the maximum at the crest for U > 15 m s
-1
. For increasing
moderate windspeeds (15 to 30 m s
-1
) the windward pre-
cipitation amounts do not change significantly (Fig. 2c),
which suggests a balance between additional windward
condensate generated upwind of the crest and the advec-
tion precipitation into the lee by the stronger winds.
A series of simulations were completed using N
m
=
0.01 s
-1
and L = 50 km, but the freezing level (FL) was
increased to 500 mb (Fig. 2c). For these higher freezing
level experiments the orographic cloud is dominated by
warm rain processes. A higher freezing level results in a
more narrow peak of precipitation over the steep wind-
ward slope compared to the lower freezing level runs
(Figs. 2a). The higher freezing level also results in a more
rapid increase in precipitation rate with increasing wind
speed between 15-25 m s
-1
, less rapid increase between
25-30 m s
-1
, and there is less precipitation spillover into
the lee.
When the freezing level is lowered to just below sea
level for N
m
=0.01 s
-1
and L=50 km (FLSFC, Fig. 2c), the
the precipitation distribution is more similar to FL750
than FL500, but for high wind speeds the FLSFC precip-
itation is less broad than FL750 since most of the snow
growth occurs at low-levels immediately over the wind-
ward slope given the very cold temperatures aloft. This
results in more snow falling out closer to the crest than
FL750.
5. ROLE OF VERTICAL WIND SHEAR ABOVE
BARRIER
Changes in the ambient winds above crest level may
also effect the precipitation distribution. Therefore, two
sensitivity runs were completed in which the winds at or
below crest level (1500 m) were initialized at 15 m s
-1
,
but either reverse or forward shear was specified above
the barrier. For the reverse shear initialization the ambi-
ent flow decreased linearly above the crest to zero at 550
mb, while for forward shear the winds increased to 30 m
s
-1
at 550 mb. These sensitivityruns were initialized with
N
m
= 0.01 s
-1
and L = 25 km.
Figure 5 shows the 6-12 h precipitation across a
1500 m barrier for the control (no shear), reverse, and
forward shear simulations. Reverse shear above the bar-
rier favors low-level mountain wave amplification and
therefore more subsidence over the crest and lee (not
shown), which in turn reduces the precipitation in this
region relative to no shear and forward shear cases.
Decreasing winds above the crest also increases precipi-
tation 50-100 km upstream of the crest through perhaps
longer time scales for ice growth aloft. In contrast, for-
ward shear favors a weaker vertically propagating grav-
ity wave (not shown), therefore reducing the upward
motion and precipitation production aloft over the wind-
ward slope. Furthermore, the stronger flow aloft with for-
ward shear also results in slightly more precipitation
advecting 10-20 km downwind of the crest.
Avg Precip (mm)
5 1015202530
0 10203040
Avg Precip (mm)
5 1015202530
0 10203040
Avg Precip (mm)
5 1015202530
0 10203040
Avg Precip (mm)
5 1015202530
0 10203040
Avg Precip (mm)
5 1015202530
0 10203040
2500 m
2000 m
1500 m
1000 m
500 m
m s
-1
Avg Precip (mm)
5 1015202530
0 10203040
Avg Precip (mm)
5 1015202530
0 10203040
Avg Precip (mm)
5 1015202530
0 10203040
Avg Precip (mm)
5 1015202530
0 10203040
Avg Precip (mm)
5 1015202530
0 10203040
2500 m
2000 m
1500 m
1000 m
500 m
m s
-1
Figure 5. The 6-12 accumulated surface precipitation across a
1500 m barrier (L=25 km) for cases with, no shear (U= 15 m s-
1), reverse shear, and forward shear above the barrier crest.
6. PRECIPITATION EFFECIENCY
The windward precipitation efficiency (PE) was cal-
culated for the region upstream (left) of the crest by sum-
ming the 6-12 h surface precipitation and dividing by the
sink terms of water vapor in the MM5 over this same
region. Figure 6 shows the PE averaged for all barrier
heights (500-2500 m) for several different N
m
, L, and FL
runs as a function of wind speed. There is a general
decrease in PE with increasing wind speed since more
hydrometeors are advected into the lee. However, for the
FL500 run, the PE remains relatively unchanged at high
windspeeds since the rain generated above the barrier
through warm rain processes falls out rapidly (Fig. 2c).
The L=25 km and N
m
=0.01 s
-1
runs have lower PEs at
higher windspeeds than simulations with larger stability
and wider terrain since weaker stability and a narrower
barrier favors more precipitation generation above the
crest (Fig. 2b), which can advect easily into the lee. The
PEs for the surface FL runs (FLSFC) drop more rapidly
with increasing wind speed (5 to 30 m s
-1
) than the
higher FL runs since more of the FLSFC precipitation is
snow that can spill into the lee. However, the PEs are still
larger in the FLSFC than the other experiments for the
same wind speed since FLSFC has a lower snow growth
region (-15
o
C 700 mb), which favors a larger percentage
of precipitation falling out over the windward slope for a
relatively wide (L=50 km) barrier (not shown). Even
though the PEs are relatively large in the FLSFC, it gen-
erates less total surface precipitation than the other
experiments since a colder atmosphere holds less mois-
ture (Figs. 2a,d). In contrast, the FL500 run favors more
flooding since not only does the atmosphere hold more
water vapor, but the lack of lateral hydrometeor advec-
tions keep the PEs relatively large at higher wind speeds.
Figure 6. The windward precipitation efficiency averaged for
all barrier heights (500-2500 m) as a function of wind speed (m
s
-1
) for the experiments labelled in the inset box.
7. CONCLUSIONS
This study uses the 2-D MM5 to illustrate some of
the sensitivities of orographic precipitation to changes in
ambient flow (U), moist static stability (N
m
), freezing
level (FL), terrain height (h
m
), and mountain half width
(L). The results suggest that the precipitation distribution
around a barrier can be strongly dependent on how the
terrain-induced gravity wave modifies the mountain cir-
culation. For example, a hydrostatic mountain wave with
a small vertical wavelength that tilts upstream with
height (favored during weak U, large N
m
, and moder-
ately wide terrain) can result in weak sinking motion a
fewkilometers above the windward slope. This keeps the
precipitation shallow and along the windward slope.
Both a narrower barrier and weaker stability favor less
tilt to the mountain wave or an evanescent wave, result-
ing in a more collapsed mountain circulation above the
crest, and more precipitation spillover (decreased precip-
itation efficiency). Reverse shear above the crest also
favorslow-levelwaveamplification and a windwardshift
in the precipitation, while forward shear favors a weaker
mountain circulation over the crest more advection into
the lee. Finally, a freezing level well above crest level
collapses the precipitation distribution around the upper-
level windward slope, with less lee side spillover, and
there is less decrease in the precipitation efficiency with
increasing wind speed.
This idealized work is currently being extended to
three dimensions using more complex terrain, with the
results are being applied to recent field programs such as
IPEX and IMPROVE.
8. ACKNOWLEDGEMENTS
This research is supported by the National Science
Foundation (Grant No. ATM-0094524). The 2-D MM5
was developed by Prof. Dave Dempsey and the MMM
Division of NCAR.
9. REFERENCES
Bruintjes, R. T., T. L. Clark, and W. D. Hall, 1994: Inter-
actions between topographic airflow and cloud/pre-
cipitation development during the passage of a
winter storm in Arizona. J. Atmos. Sci., 51, 48-67.
Durran, D. R., and J. B. Klemp, 1983: A compressible
model for the simulation of moist mountain waves.
Mon.Wea. Rev., 111, 2341-2351.
Jiang, Q, and R.B. Smith, 2002: Microphysical times-
cales and orographic precipitation. Accepted to J.
Atmos. Sci.
Klemp, J. B., and D. R. Durran, 1982: An upper bound-
ary condition permitting internal gravity wave radi-
ation in numerical mesoscale models. Mon. Wea.
Rev., 111, 430-444
Neiman, P.J., and co-authors 2002: The statistical rela-
tionship between upslope flow and rainfall in the
California’s coastal mountains: observations during
CALJET, accepted to Mon. Wea. Rev.
Pandley, G.R., and co-authors, 1999: Precipitation struc-
ture in the Sierra Nevada of California during win-
ter. J. Geophys. Res., 104, 12019-12030.
Sinclair, M.R.,1994: A diagnostic model for estimating
orographic precipitation. J. Appl. Meteor, 33, 1163-
1175.
50 100 150 200 250 300 350
0 5 15 25
50 100 150 200 250 300 350
0 5 15 25
50 100 150 200 250 300 350
0 5 15 25
REV SHEAR
FWD SHEAR
NO SHEAR
km
mm
Precip Efficiency
5 1015202530
0.0 0.2 0.4 0.6 0.8 1.
0
Precip Efficiency
5 1015202530
0.0 0.2 0.4 0.6 0.8 1.
0
Precip Efficiency
5 1015202530
0.0 0.2 0.4 0.6 0.8 1.
0
Precip Efficiency
5 1015202530
0.0 0.2 0.4 0.6 0.8 1.
0
Precip Efficiency
5 1015202530
0.0 0.2 0.4 0.6 0.8 1.
0
Precip Efficiency
5 1015202530
0.0 0.2 0.4 0.6 0.8 1.
0
N=.01 L=50km
N=.005 L=50km
N=.01 L=25km
N=.005 L25km
N=.01FL500 L=50km
N=.01FLSFC L=50km
15 20 25 30105 m s
-
1
1
2
3
4
5
6
6
5
1
4
3
2
