On the Moisture Extraction of Mountains via Deep Convection

60 40 20 0 20T in ◦ C

0.5

5.0

10.0

12.5

22.0

z i

n k

m Tropopause he ight

T

RH

0 2 4 6 8 10 12 14

u in m /s

0.5

5.0

10.0

12.5

22.0

U5 U1 0U0

10 20 30 40 50 60 70RH in %

10 20 30 40 50 60 70RH in %

r in km

he

igh

t in

m

(a )

r in km

(b)

r in km

(c)

r in km

(d)

Isolated Mountain

Height scales H={250,500,750,1000, 1500 m}Width scales A={5,7,10,14,20,25,30 km}

-

Origin and Breakdown of P~V Scaling

Full range of simulations

Deep convection is the dominant mode of summer precipitation in many extratropical regions. Especially during episodes of weak synoptic forcing, mountains are deep convection hot spots, in particular due to the elevated moistening by upslope flows. What is the relative contribution of different orographic scales on deep convective precipitation? Or alternatively: which of the following mountains extracts most rain via deep convection?

Constant lapse rate -7 K/km, RH decreases from 70 % in PBL to 40 % at tropopause. Experiments with initially resting atmosphere (U0) and with sheared background wind (U5 and U10) are performed.CAPE < 1400 J/kg

... but we find a breakdown of the scaling for large-scale flow past the tallest mountains (>1000 m)

The breakdown of the scaling is likely related to the venting of the mountaintop heat anomaly by the large-scale flow (c.f. Nugent et al. 2014). Linear theory (Smith and Lin, 1982) implies that the relative contribution of the thermal component to the updraft speed W decreases with background wind speed scale U as:

W~1/U²

10 ensemble members per configuration (differ only in initial noise in lowest T level)5 days free run per configuration VARh: keep a=14 km fixed and vary full range of heights h. VARa: keep h=500 m fixed and vary full range of widths a.

Col

lisio

n be

twee

n c

old

pool

an

d up

slop

e

flow

(do

wns

trea

m p

ropa

gat

ion

of c

ollis

ion

poin

t)T

ime

Moi

sten

ing

is r

ela

tive

to t=

06:0

0 (b

efor

e su

nris

e)

Surface latent heat fluxes (contours in W/m²)hardly contribute to moisture excess over the mountain

0

1

2

3

4

5

z in

km

9 :0 0

h250a209 :0 0

h500a14

1ms

9 :0 0

h1000a10

0

1

2

3

4

5

z in

km

1 1 :0 0 1 1 :0 0

1ms

1 1 :0 0

0 20 40

r in km

0

1

2

3

4

5

z in

km

1 3 :0 0

0 20 40

r in km

1 3 :0 0

1ms

0 20 40

r in km

1 3 :0 0

0.25

0.50

0.75

1.00

1.25

1.50

1.75

2.00

∆q v

in g

/kg

6:00

9:00

12:00

15:00

18:00

loca

l ti

me

h2 5 0 a20

0.1

0.1

0.10.1

0.5

1.0

2.0 4.0

h5 0 0 a14

0.1

0.1

0.5

0.5

0.5

1.02.0

4.0

8.0

h1 0 0 0 a1 0

0.1

0.5

0.5

1.0

2.0

4.0

8.0

6:00

9:00

12:00

15:00

18:00

loca

l ti

me

h2 5 0 a20

0.1

0.1

0.1 0.1

0.1

0.51.0

2.04.0

h5 0 0 a14

0.1

0.1

0.5

0.5

0.5

1.02.0

4.08

.0

h1 0 0 0 a1 0

0.1

0.5

0.5

1.0

2.0

4.0

8.0

0 20 40r in km

6:00

9:00

12:00

15:00

18:00

loca

l ti

me

40.080.0

120.0

120.0

160.0

160.0

200.0

200.0

240.0

0 20 40r in km

40.080.0

120.0

120.0

16

0.0

160.0

200.0240.0

0 20 40r in km

40.0 40.080.0

120.0

120.0

16

0.0

160.0

200.0

240.0

6.43.21.60.80.4

0.40.81.63.26.4

u i

n m

/s

1.60.80.40.20.1

0.10.20.40.81.6

w i

n m

/s

20242832364044

TW

P i

n m

m

cold pools

surface rain rate

in mm/h

The P~V scaling is surprising given that the deep convective life cycle over mountains with the same volume V strongly differs in terms of timing and amplitude. E.g. the moistening of the atmosphere by the upslope flows is determined by the slope of the mountain and not the volume:

Deep Convective Life Cycle at Mountains with the Same Volume

Hovmöller diagrams of radial wind u, vertical wind at 2 km AGL, and total water path (TWP)further highlight these differences:

1 2 3 4 5

day

0

1

2

3

4

5

6

7

mm

/da

y

0

1

2

3

4

5

6

7

h250a20h500a14h1000a10

h250a20

h500a14

h1000a10

In the statistical mean (ensemble and day average) the rain amount P (at r<30 km) is fully determined by the mountain volume V across the whole range of simulations:

P0=2.9 mm/ day, V0 =458 km³(rain amount and volume for h1000a10 mountain)

0 2 4 6 8 10 12 14

0

1

2

3

4

5

6

7

P/P0

250 m

500 m

750 m

1000 m

1500 m

0.0 0.5 1.0 1.5 2.0V/V0

0.0

0.5

1.0

1.5

2.0

P/P0

Mo

un

tain

Rai

n A

mo

un

t in

PO

Mountain Volume in VO

Mo

un

tain

Rai

n A

mo

un

t in

PO

day

Mo

un

tain

Rai

n A

mo

un

t in

mm

/day

Mountain Volume V ControlsRain Amount P: the P~V Scaling

The P~V scaling emerges only statistically: The daily rain amountP at r<30 km for three mountains with the same volume (h250a20,h500a14 and h1000a10) undergoes considerable daily and ensemblespread:

We use the Consortium for Small-Scale Modeling (COSMO, Baldauf et al., 2011) in climate mode (CCLM) at a Δx=1 km horizontal grid spacing. Full physics parameterizations are employed. No shallow/deep convection parameterization.

512x512 km² domain, 22 km deep, double periodic boundaries, no Coriolis force, extratropical summer insolation, COSMO is interactively coupled to the 2nd generation, multi-layer soil model TERRA_ML. (c.f. Imamovic et al., 2017).Idealized atmospheric profiles:

The P~V scaling is also valid for cos² mountain profile and experiments with multiple Gaussian peaks

- We found a linear scaling between rain amount P and mountain volume V- The P~V scaling regime is valid for a surprisingly large portion of parameter space. The scaling does not seem to depend on the details of mountain profiles / number of peaks.- The scaling prevails if background wind is present for mountains lower than 750 m. For strong background wind flowing past tall mountains the scaling breaks down, likely due to the ventilation of the mountaintop heat anomaly by the background flow.

Impact of Large-Scale Flow on the P ~ V scaling

Adel Imamovic, Linda Schlemmer, and Christoph Schär

Motivation and Research Q

Numerical Experiments

Conclusions

Acknowledgments References

-3 -1 1 3

-3

-1

1

3

y/a

h5 0 0 a14 _U0

-3 -1 1 3

-3

-1

1

3

h5 0 0 a14 _U5

-1 1 3

-3

-1

1

3

h5 0 0 a14 _U1 0

1ms

-3 -1 1 3

x/a

-3

-1

1

3

y/a

h1 5 0 0 a1 4 _U0

-3 -1 1 3

x/a

-3

-1

1

3

h1 5 0 0 a1 4 _U5

-1 1 3

x/a

-3

-1

1

3

h1 5 0 0 a1 4 _U1 0

1

5

10

15

20

25

30

rain

am

ou

nt

in m

m/d

1ms

analysis region

P~V scaling prevails with large-scale flow for mountains < 750 m in height and all widths....

We thank PRACE for awarding us acces to Piz Daint , the Cray XC-50 system at the CSCS, the Swiss National Supercomputing Centre in Lugano, Switzerland. Special thanks are extended to the CLM community for providing access to the COMSO model.

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5V/V0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

P/P0

Gauss

cos2

m ult iplepeaks

Baldauf et al. (2011): Operational convective-scale NWP with the COSMO model: description and sensitivities, JAS.Imamovic, Schlemmer, and Schär (2017): Collective impacts of orography and soil moisture on the soil moisture-precipitation feedback, GRL.Imamovic, Schlemmer, and Schär: On the moisture extraction of mountains via deep convection, in prep.Nugent, Smith, and Minder (2014): Wind speed control of tropical orographic convection, JAS. Smith and Lin (1982): The addition of heat to stratified air-stream with application to the dynamics of orographic rain, QJRMS.

horizontal wind field at 500 m height

Accumulated rain amount for the h500a14 and h1500a14as a function of background wind (U0,U5, and U10):