Long-Term Vertical Velocity Statistics Derived from Doppler Lidar in the Continental Convective Boundary Layer

June 16, 2016 1

LARRY BERG1, ROB NEWSOM1, DAVE TURNER2, RAJ RAI1

1Pacific Northwest National Laboratory 2NOAA National Severe Stoms Laboratory

Contact: larry.berg@pnnl.gov

!   Distributions of vertical velocity (w) are needed to understand both PBL turbulence and boundary-layer clouds

!   Most studies of w statistics are derived from short-term aircraft observations or tower based studies !   BAO (300 m) ! Cabaw (213 m)

!   Long-term measurements of w statistics are lacking, but are needed for evaluating model results and developing new parameterizations !   ARM LASSO (LES ARM

Symbiotic Simulation and Observation) project

!   Data collected by Department of Energy’s Atmospheric Radiation Measurement (ARM) Climate Research Facility provide a unique opportunity to address this need.

24 Rai and Berg et al.

data from a single point at the center of domain), but not resolved in both time and626

space. Vertical profiles of wind speed and potential temperature have been prepared627

for the seven spatial locations within D06 for half-hour period 1300-1330. The loca-628

tion of SW and S are exactly due south of location W and C, whereas the location629

of N and NE are due north of the C and E. These four additional locations (SW,630

N, E, and S) are located approximately 660 m inside the south and north boundary631

of domain D06. Figure 17a shows the TKE profile at the seven locations; the rela-632

tive locations are shown in the inset box. Note that the solid-lines, dashed-line, and633

dotted-lines represent the region with the higher elevation (W (544 m), SW (506 m),634

S (555 m), and C (443 m)), center point, and lower elevation (N (305 m), E (318 m),635

and NE (226 m)) of the domain D06. Similarly, the legend lines with red, blue, and636

black represent the locations that are generally aligned with the predominant flow637

(i.e., southwesterly). The TKE profile shows a great deal of variability between these638

seven locations. In general, the TKE observed at the lower elevations is smaller than639

the TKE at higher elevations. However, the TKE profile between the two consecutive640

locations along the mean flow (i.e., E and S; N and W) are comparable except be-641

tween the two locations NE and C. The smallest TKE value observed for location NE642

may be due to its relatively large offset distance from the wake region (refer to Fig.643

7). It is interesting to note that the maximum TKE at locations C and N are found644

close to the surface. Figure 17b shows the standard deviation sw

profile for the ver-645

tical wind component at all locations. The shape of the profile is almost similar to646

that of the TKE profile and peak values of sw

can be found anywhere from 300 to647

600 m above the surface. Given the relative shape of the TKE and sw

profiles, the648

low-altitude maximum TKE at points C and N is most likely due to variations in the649

horizontal wind components. The mean horizontal and vertical wind components are650

shown in Fig. 17c and Fig. 17d, and the figure highlights the variability occurring in651

the seven locations. The horizontal wind is systematically slower in the locations hav-652

ing lower elevation (dotted lines) and exhibit smaller vertical wind shear compared653

������ ��� ����

0 1.5 3

���

��

0.6

1.2

1.8 a)

�� �� ���0 0.9 1.8

b)

� �� ���-0.7 0 0.7

c)

��� � ��� �� ���

0.5 2.5 4.5

d)

��� � ��

0 0.15 0.3

e)

SW S

W C E

N NE

Fig. 17 A half-hour (1300-1330 PST) averaged profile of a) TKE, b) standard deviation of vertical compo-nent of wind, c) mean of vertical component of wind, d) horizontal wind speed, and e) variance of potentialtemperature at seven locations SW, S, W, C, E, N, and NE. The green box in the inset shows the sevenlocations in domain D06.

Motivation

June 16, 2016 2

TKE  and  σw  generated  from  LES  of  region  of  complex  terrain.    

Outline  •  Data  descrip:on  •  Composites  of  w  stats  •  Sor:ng  the  results  •  Conclusions  

ARM Doppler Lidars and Analysis Criteria

!   Halo-Photonics DL Deployed at ARM fixed sites and AMF !   Near-IR (1.5 µm) !   Range is typically less than 2 km for clear-air

retrievals !   Value Added Product (VAP) has been

developed that includes wind profiles and key statistics: variance, skewness, and kurtosis

!   Utilize ARM meteorological and flux data, wind profiles from 915 MHz radar wind profilers

!   Data selection criteria !   Require SNR to be ≥ 0.03 (more stringent

criteria used for higher order stats) !   Define zi using a threshold of variance

(0.04—based on Tucker et al. 2009) ! zi > 0, w* > 0 (implies positive surface heat flux) !   Cloud Fraction < 0.001

June 16, 2016 3

50

45

40

35

30

25

-120 -110 -100 -90 -80 -70

ARM SGP Site

Representative Case: 18 July 2015

June 16, 2016 4

2.5

2.0

1.5

1.0

0.5

0.0

z or

zi (

km)

12:007/18/15

16:00 20:00 00:007/19/15

Date and Time (UTC)

3.5

3.0

2.5

2.0

1.5

1.0

0.5

0.0

Heat Flux (K m s -1) or w

* (m s -1)

2.01.51.00.50.0

σw

2 (m2 s -2)

ECOR w* ECOR Sensible Heat Flux zi

ECOR=Eddy  Covariance  System  

Composite Statistics

June 16, 2016 5 12:00 15:00 18:00 21:00 00:00

Time (UTC)

2.5

2.0

1.5

1.0

0.5

0.0

z (k

m)

2.5

2.0

1.5

1.0

0.5

0.0

z (k

m)

2.5

2.0

1.5

1.0

0.5

0.0

z (k

m)

Skewness

Kurtosis

1.21.00.80.60.40.20.0σw

2

1.21.00.80.60.40.20.0

43210Kurtosis  of  Gaussian  Distribu:on  =  3  

Day:mes  values  posi:vely  skewed  

Peak  value  smeared  by  averaging  

More  peaked  

Less  peaked  

Sensitivity of Statistics to Parameters

!   The w statistics can be sorted by many different ways—think of factors that could influence the PBL turbulence !   Time of day !   Wind direction !   Season !   Surface shear stress/friction velocity (u*) !   Wind Shear at PBL top !   Static stability

!   Use Kolomogorov-Smirnov test to determine if differences are statistically significant !   Look to see if values could be from the same parent distribution

June 16, 2016 6

Change in Variance Profile with Time

!   All values of σw2 have been

normalized by w*

!   Values tend to be smaller than short-term results presented by Lenschow et al. (2012):

!   Normalization works best in late-morning and afternoon !   PBL approximately steady

state

June 16, 2016 7

0.50.40.30.20.10.0

Median σw2/w*

2

1.2

1.0

0.8

0.6

0.4

0.2

0.0

z/z i

18:00

Time (UTC)

14:30

22:30

2015 All Months, Southerly Winds

Limit  analysis  to  17  to  23  UTC  

€

σw2

w∗2 =1.8 z

zi

$

% &

'

( )

2 3

1− 0.8 zzi

$

% &

'

( )

2

Sensitivity of σw2/w*

2 to Wind Direction

!   Differences in land use and surface roughness associated with wind direction could lead to differences in velocity statistics

June 16, 2016 8

1.2

1.0

0.8

0.6

0.4

0.2

0.0

z/z i

0.50.40.30.20.10.0σw

2/w*2

5000# of Obs.

Easterly Southerly Westerly

Limit  analysis  to  periods  with  southerly  winds  

Error  bars:  75th  and  25th  percen:les  

ARM  CF  

Dependence on Season

June 16, 2016 9

σw2/w*

2  changes  with  season,  differences  in  surface  heterogeneity?  Skewness  larger  in  PLB  top  during  warm  seasons    

1.2

1.0

0.8

0.6

0.4

0.2

0.0

z/z i

0.60.50.40.30.20.10.0σw

2/w*2

0.80.40.0Skewness

65432Kurtosis

DJF MAM JJA SON

Dependence on u*

!   Values of u* from surface flux measurements were sorted to determine critical values !   u* in Lenschow et al. (2012) ranged from 0.16 to 0.52 ms-1

June 16, 2016 10

1.2

1.0

0.8

0.6

0.4

0.2

0.0

z/z i

0.60.50.40.30.20.10.0

σw2/w*

21.00.80.60.40.20.0-0.2

Skewness65432

Kurtosis

Large u* Small u*

Large  values  of  u*  lead  to  small  values  of  σw2/w*

2  and  kurtosis  

Large:  0.59  ms-­‐1  Small:  0.27  ms-­‐1  

Dependence on Stability

!   Stability defined using –zi/L !   Values greater than 30 are unstable, less than 30 moderately unstable

June 16, 2016 11

1.2

1.0

0.8

0.6

0.4

0.2

0.0

z/z i

0.60.50.40.30.20.10.0

σw2/w*

21.00.80.60.40.20.0-0.2

Skewness65432

Kurtosis

Unstable Moderately Unstable

Unstable  cases  have  smaller  values  of  σw2/w*  and  kurtosis  

Dependence on Wind Shear Across the Boundary-Layer Top

!   Wind shear determined from radar wind profiler !   Based on wind speed differences between z/zi of 1.1 and 0.9

June 16, 2016 12

1.2

1.0

0.8

0.6

0.4

0.2

0.0

z/z i

0.60.50.40.30.20.10.0

σw2/w*

21.00.80.60.40.20.0-0.2

Skewness65432

Kurtosis

Large Shear Small Shear

Large:  1.4  ms-­‐1  Small:  -­‐0.6  ms-­‐1  

Cases  with  large  amounts  of  shear  are  less  skewed  over  much  of  the  BL  

Conclusions

!   Data from ARM Doppler lidars provide a unique opportunity to look at long-term vertical velocity statistics

!   Scaling with w* works best for cases when BL is quasi-stationary !   There as systematic differences in the σw

2/w*2, skewness, and

kurtosis associated with differences is: !   Season !   Surface stress: smaller

values of u* lead to large values of σw

2/w*2

!   Stability: Moderately unstable conditions lead to larger values of σw

2/w*2 and kurtosis

!   Wind Shear across the BL top: small values of shear lead to larger values of skewness

!   Future efforts will extend work to stable conditions

June 16, 2016 13

Less  intense  mixing  at  surface  leads  to  larger  variance  

Less  intense  mixing  at  boundary-­‐layer  top  leads  to  larger  skewness  

Acknowledgments:  This  work  was  supported  by  the  DOE  Atmospheric  System  Research  (ASR)  Program  and  Atmospheric  Radia:on  Climate  Research  Facility  

Variance  largest  in  spring;  Skewness  larger  in  warmer  seasons  

June 16, 2016 14

3-Line (or more) Header for New PNNL PowerPoint Template / Full-Color Background (if supported by content)