Leipzig, Germany, September 14th-20th, 2003
Raman Lidar measurements of atmospheric temperature
during the International H2O Project
Paolo Di Girolamoa, Rocco Marchesea,
David N. Whitemanb, Belay B. Demozb
aDIFA,
Università degli Studi della Basilicata,
C.da Macchia Romana, 85100 Potenza, Italy
bNASA/GSFC, Mesoscale Atmospheric Processes Branch,
Greenbelt, MD 20771, USA
Comprehension
of meteorological
processes and
climate trends
Measurements
of
atmospheric
temperature:
• High accuracy
• High time and space resolution
• Global coverage.
Observational requirements for networks of groundbased and satellite remote sensors
• World Meteorological Organization (WMO)
Leipzig, Germany, September 14th-20th, 2003
Measurements of atmospheric
temperature:
Vertical extent = up to the LS
Accuracy = 0.7 K
Vertical resolution = 0.1 km
Temporal resolution = 15 min
Globally distributed
Lidar systems have the potential to achieve these obs requirements
Lidar measurements of atmospheric temperature:
• Combined Rayleigh-vibrational Raman scattering technique
(Hauchecorne and Chanin, 1980)
• Differential absorption technique
(Mégie, 1980; Theopold and Bösenberg, 1993)
• Rayleigh backscatter spectral width measurement technique
(Fiocco et al., 1971)
• Pure rotational Raman technique
(Cooney, 1972)
Measurements reported in this presentation make use of the pure
rotational Raman (RR) technique in the UV region.
All pure rotational Raman lidar measurements reported in literature
have been performed in the visible domain.
Leipzig, Germany, September 14th-20th, 2003
0-16 km
UV
T
 0.5  T
Advantages of the
use of UV laser
light instead of
visible
0-16 km
VIS
  
4
 UV   VIS
• increase the precision
• achieve better daytime performances
due to reduced sky background
• safer in terms of hazard for eye injury
Threshold for thermal retinal damage:
• 3 orders of magnitude lower than in the visible
Maximum allowed exposition of human eye (for laser pulses 1 to
100 ns in duration):
•
30 J/m2 in the spectral region 280-400 nm
•
5 mJ/m2 in the 400-700 nm region (EN 60825-1, 2001).
UV laser beams used in most lidar applications result to be eye-safe
within few hundred meters from the laser source
Leipzig, Germany, September 14th-20th, 2003
NASA Scanning Raman Lidar, SRL
• Mobile system in a environmentally
controlled trailer
• Nd:YAG laser
• 0.76 meter telescope
• Large aperture scanning mirror
• Outfitted with a UV rotational Raman
temperature measurement capability
prior to IHOP
Nd:YAG laser:
Single pulse energy @ 354.7 nm = 300 mJ
Pulse repetition rate = 30 Hz
Linewidth (FWHM)=1 cm-1
Frequency stability<0.5 cm-1
Beam divergence=250 mrad
Leipzig, Germany, September 14th-20th, 2003
Filter assembly:
based on interference filters (IFs)
Sensitivity study
• careful analysis of the temperature dependence of rotational lines
• maximising measurement precision
• maximising measurement sensitivity
• minimize potential sources of contamination, as RR scattering from
water vapour
1E-32
40
High J filter
Low J filter
T=310 K
T=280 K
-1
30
2
25
1E-34
20
15
1E-35
10
5
1E-36
351
352
353
354
355
Wavelength (nm)
356
357
Filters specs
0
358
Filter transmission (%)
Cross Section (m sr )
1E-33
35
Filter assembly
Low-J High-J
CWL (nm)
354.3
352.9
0.2
1.0
CWL transmission
30 %
30 %
Blocking @ 355 nm
10-6
10-6
FWHM (nm)
International H2O Project (IHOP)
Location: Southern Great Plains (USA)
Period: May-June 2002
Main goal: study the role of water
vapour in convection initiation and
to improve on quantitative
precipitation forecasting (QPF).
SRL deployment: Homestead site,
Western Oklahoma
• SRL operated for approximately 35 days during IHOP
• Most of the measurements were carried out in vertically pointing mode
Approx. 200 hours of SRL data
Radiosonde launch station next to SRL
148 radiosondes launches
Leipzig, Germany, September 14th-20th, 2003
CALIBRATION FUNCTION
PhiJ  zT 
RT  
 exp a T  b 
PloJ  zT 
a
T
ln R  b
• exactly valid for two individual lines (Arshinov et al., 1983)
• can be assumed valid also for portions of RR spectrum
Low-J filter: 4 rotational lines
High-J filter: 17 rotational lines
J=5 from O2,
J=7 from O2,
J=4 from N2,
J=5 from N2,
J=14-23 from O2,
J=19-32 from O2
Systematic error assuming calibration analytical expression valid for
portions of RR spectrum < 1.5 K.
Leipzig, Germany, September 14th-20th, 2003
Calibration constants a and b determined through comparison with
simultaneous radiosondes.
• 6 lidar-radiosonde comparisons
• Inclusion of both night-time and twilight cases
• a = -758 ± 6 and b = 0.95 ± 0.02
Systematic error associated with indetermination
of calibration constants
max 2 K
Systematic error associated different overlap
functions in the two RR channels
• near range, < 1-2km
max 2 K
Systematic error associated with laser frequency
looking accuracy/stability < 0.5 cm-1
max 0.5 K
Assuming the different
sources of systematic
error to be independent
Overall systematic error max 3 K z < 2 km
max 2 K z > 2 km
25
MEASUREMENTS
NIGHT TIME
MEASUREMENT
• Error bars statistical uncertainty only
• Ended 1/2 hour before sunrise
• almost clear sky conditions
Rand.err.  1.5 K at 15 km
Height (km)
Lidar measurements
• up to approx. 23 km (rand. error > 5 K)
20
Lidar-radiosonde comparison
15
9 June 2002
10
• Good agreement
• Deviations
 < 2 K up to 14 km
5
 < 3 K up to 17 km (max. sonde height)
• Average bias = 0.5 K
• RMS deviation = 1.2 K
Sonde (10:46 UT)
Lidar (09:26-10:55)
0
180 200 220 240 260 280 300
Leipzig, Germany, September 14th-20th, 2003
Temperature (K)
-2 0 2
T
14
MEASUREMENTS
• started 1 hour before sunset (twilight
12
conditions)
• almost clear sky conditions
TWILIGHT
MEASUREMENT
2 June 2002
Lidar measurements
• up to approx. 14 km
• smaller vertical extent vs night-time
 day-dusk transition (lidar
performances degraded by
solar background noise
Height (km)
10
8
6
4
Lidar-radiosonde comparison
• Good agreement
• Average bias = 0.2 K
• RMS deviation = 1.8 K
2
Sonde (02:40 UT)
Lidar (01:00-02:30)
0
180 200 220 240 260 280 300
Leipzig, Germany, September 14th-20th, 2003
Temperature (K)
-2 0 2
T
Simulations → quantify measurement precision of RR technique
• 355 and 532 nm
• nigh-time and daytime operation
T z 
PloJ z   bkloJ PhiJ z   bkhiJ
T z  
R z 

2
2
z 
R
PloJ z 
PhiJ
Behrendt and Reichardt, 2000
• Poisson statistics for backscatter and background signals
• Pressure, temperature and humidity from US standard atm (1976)
• Aerosol extinction data from the ESA ARMA (1999), median model
• No clouds
Filters specs at 532 nm were
Overall rec. efficiency @ 355 nm=0.055
defined in order to isolate the same
• receiving optics reflectivity (0.9)
rotational Raman lines as at 355
• filter transmission (0.3)
nm (same quantum numbers).
• detector quantum efficiency(0.2))
Overall rec. efficiency @ 532 nm=0.055
•receiving optics reflectivity (0.9)
•filter transmission (0.5)
•detector quantum efficiency(0.12))
Leipzig, Germany, September 14th-20th, 2003
Same power-aperture product @
355 and 532 nm, as SRL
Daylight background
• Mainly due to scattering of sunlight
• Determined from Modtran database
• Sun zenith angle = 400
•
bk355=0.15 bk532
T for 15 minute integration
Daytime simulations
Two spectral selection configurations:
• use of IFs only
• use of combination of a Fabry-Perot
interferometer and IFs
to reduce sky background (Arshinov et
1E-4
70000
60000
Vertical resolution = 100 m
(to fit WMO requirements)
Night-time
T355 < T532 20-50 %
t=1 h
t=15 min
T < 0.4 K
T < 0.7 K
z<15 km
@ 355 and 532 nm
50000
Height (m)
between adjacent lines (3.3 cm-1) and the
spectral width of individual lines (0.05 cm-1).
0.01
0.1
1
10
100
1000
SIMULATION
al., 2001; Bobrovnikov et al., 2002)
Gain in signal-to-background ratio 
65 (ratio between average separation
1E-3
T night, 532 nm
T day, 532 nm
T day, 532 nm high res
T night, 355 nm
T day, 355 nm
T day, 355 nm high res
40000
30000
20000
10000
Satisfies target
observational
0
1E-4 1E-3 0.01
0.1
1
10
requirementsT for 1 hour integration
from WMO
100
1000
Day-time
T for 15 minute integration
1E-4
70000
IFs only
1E-3
0.01
0.1
1
10
100
1000
SIMULATION
T355  0.2 T532
t=1 h, z<15 km
60000
T355 < 2.5 K
50000
T355  0.2 T532
t=1 h, z<15 km
T355 < 0.7 K
Height (m)
Fabry-Perot + IFs
T night, 532 nm
T day, 532 nm
T day, 532 nm high res
T night, 355 nm
T day, 355 nm
T day, 355 nm high res
40000
30000
20000
10000
0
1E-4
Leipzig, Germany, September 14th-20th, 2003
1E-3
0.01
0.1
1
10
T for 1 hour integration
100
1000
Conclusions and Future Plans
• Measurements of atmospheric temperature in the UV have been performed based
on the application of the pure rotational Raman technique
• First successful attempt to perform RR temperature measurements throughout the
troposphere in the UV region
 eye-safe concerns are less stringent than VIS and IR
 increase the precision
 better daytime performances due to reduced sky background
• Simulation have been performed in order to quantify the potentialities in terms of
measurement precision of the RR lidar technique both in the visible and UV
• Simulations reveal that night-time measurements satisfy target observational
requirements from WMO
Future:Implement high resolution spectral detection based on FP+IFs
1E-32
40
High J filter
Low J filter
T=310 K
T=280 K
-1
30
2
25
1E-34
20
15
1E-35
10
5
1E-36
351
352
353
354
355
Wavelength (nm)
356
357
0
358
Filter transmission (%)
Cross Section (m sr )
1E-33
35
Photomultipliers:
• included inside unshielded housings
• performances altered by the laser induced electromagnetic
noise (SIN)
signal discrimination
level for photon
counting increased
(2mV → 3 mV)
Leipzig, Germany, September 14th-20th, 2003
• reduction in photon count rates
(both low-J and high-J chns)
• system configuration
unoptimized for temperature
measurements.
BACK-UP SLIDES
16.000
14.000
temperatura_lidar
sonda step
Serie3
12.000
10.000
8.000
6.000
4.000
2.000
0.000
180.0
200.0
220.0
240.0
260.0
280.0
300.0
320.0
340.0
Cirrus cloud between 10.5-12 km
Peak scattering ratio = 10
Raw lidar data vertical resolution = 30 m
Vertically smoothing = 600 m in order to reduce signal statistical fluctuations
Smoothing procedure
• binning
• assigning equal weight to each data point
Leipzig, Germany, September 14th-20th, 2003
LASER
Nd:YAG laser:
Single pulse energy @ 354.7 nm = 350 mJ
Pulse repetition rate = 30 Hz
Unseeded
Linewidth (FWHM)=1 cm-1
Frequency stability=0.5 cm-1
Beam diverengence=250 mrad
According to manufacturer specifications, laser fluctuations resulting
from thermal drifts inside the laser cavity are expected to guarantee a
frequency looking accuracy/stability better than of 0.5 cm-1. 3.1GHz/K
Consequent changes in amplitude of detected signals, primarily the
high J signal, may lead to a systematic error which has been
estimated to not exceed 0.5 K.
FILTER ASSEMBLY
A filter blocking at the laser wavelength of 10-6 has been estimates
to prevent from contamination due to elastic echoes from
aerosol/cloud structures with a scattering ratio up to 10.
 Gain in signal-to-background ratio @ 65 (ratio between
average separation between adjacent lines (3.3 cm-1)
and the spectral width of individual lines (0.5 cm-1).
Requirement
Hor. Res.
(km)
Vert. Res.
(km)
Obs. Cycle
(h)
RMS (K)
User
Application
Lower troposph.
10/500
0.3/3
0.5/12
0.5/3
WMO
Regional NWP
Upper troposph.
10/500
1/3
0.5/12
0.5/3
WMO
Regional NWP
Lower tratosph.
10/500
1/3
0.5/12
0.5/3
WMO
Regional NWP
Lower troposph.
50/500
0.3/3
1/12
0.5/3
WMO
Global NWP
Upper troposph.
50/500
1/3
1/12
0.5/3
WMO
Global NWP
Lower tratosph.
50/500
1/3
1/12
0.5/3
WMO
Global NWP
Lower troposph.
20/200
0.1/2
3/12
0.5/3
WMO
Synopt. Meteor.
Upper troposph.
20/200
0.1/2
3/12
0.5/3
WMO
Synopt. Meteor.
Lower tratosph.
20/200
0.1/2
3/12
0.5/3
WMO
Synopt. Meteor.
Lower troposph.
5/200
0.5/1
0.25/1
0.5/2
WMO
Nowcasting
Upper troposph.
5/200
1/3
0.25/1
1/2
WMO
Nowcasting
Lower troposph.
50/500
0.5/2
6/72
0.5/1
WCRP
SPARC
Upper troposph.
50/500
0.5/2
6/72
0.5/1
WCRP
SPARC
Lower tratosph.
50/500
0.5/2
6/72
0.5/1
WCRP
SPARC
Lower troposph.
50/500
0.3/3
3/12
0.5/3
WCRP
Global Modelling
Upper troposph.
50/500
1/3
3/12
0.5/3
WCRP
Global Modelling
Lower tratosph.
50/500
1/3
3/12
0.5/3
WCRP
Global Modelling
Leipzig, Germany, September 14th-20th, 2003
Single attempt to use the RR
technique in the UV region:
• Agnew and Twort (2002)
Refractivity measurements
up to 3 km obtained from the
combination
of
RR
temperature measurements
with vibrational Raman water
vapour measurements.
Measured parameters
• water vapor mixing ratio
• particle extinction, backscattering and depolarization
• cloud liquid water, cloud droplet radius and number density
Exclusively during IHOP: temperature profile through the rotational
Raman technique
International H2O Project (IHOP)
Location: Southern Great Plains (USA)
Period: May-June 2002
Main goal: study the role of water
vapour in convection initiation and
to improve on quantitative
precipitation forecasting (QPF).
SRL deployment: Homestead site,
Western Oklahoma
• SRL operated for approximately 35 days during IHOP
• Most of the measurements were carried out in vertically pointing mode
Leipzig, Germany, September 14th-20th, 2003
• SRL outfitted with a UV rotational Raman temperature measurement
capability prior to the field campaign.
• Filter assembly developed at University of Basilicata.
• Filters’ specifications resulted of a detailed sensitivity study
• based on a careful analysis of the temperature dependence of
rotational lines, considering different temperature regimes;
• maximising measurement precision.
12.000
10.000
average over 60 bins 7.5 m each
8.000
HEIGHT
6.000
sonda_step
temperatura_lidar
sonda
4.000
2.000
0.000
100.0
150.0
200.0
250.0
TEMPERATURE
300.0
350.0
Target requirements from:
•
•
•
•
•
•
Global NWP
Regional NWP
Synoptic meteorology
Nowcasting,
Global climate modelling
SPARC
First successful attempt to perform RR temperature measurements in
the UV throughout the troposphere.
The use of the alternative the analytical expression

RT   exp a' T 2  b' / T  c'
leads to slightly smaller systematic errors not exceeding 1 K

Lidar-radiosonde comparison
• Good agreement
• Deviations
 <3K
• Average bias = 0.2 K
• RMS deviation = 1.8 K
 Larger RMS vs night-time due to
larger statistical uncertainty
Night-time measurements
lidar up to approx. 23 km
• Lidar-radiosonde comparions
 Deviations < 2 K up to 14 km
 Average bias = 0.5 K
 RMS deviation = 1.2 K
Night-time
T355 < T532 20-50 %
t=1 h, z<15 km
• T ,T < 0.4 K
Day-time
IFs only
T355  0.2 T532
t=1 h, z<15 km
Day-time
Fabry-Perot+IFs
T355  0.2 T532
t=1 h, z<15 km