Hyperspectral Sounding with AIRS: New Views of our Atmosphere

1
Hyperspectral Sounding with AIRS New Views of
our Atmosphere

• Atmospheric Spectroscopy Laboratory
• L. Larrabee Strow
• Scott Hannon
• Howard Motteler
• Sergio DeSouza-Machado
• Ji Gou
• Yvonne Edmonds (now at Stanford Univ.)
• Eric Maddy (with Chris Barnet)
• Simon Carn (JCET)
• Collaboration with Prof. Mcmillan and his group
• AIRS Science Team
• H. H. Aumann, NASA-JPL
• Dave Tobin, University of Wisconsin
• Chris Barnet, NOAA NESDIS

2
Overview

• Emphasis on climate applications
• AIRS developed primarily for weather forecasting
• 1st new weather satellite technology in 20 years
• AIRS instrument the first hyper(ultra?)spectral
  sounder
• Physics of the instrument
• Radiative transfer
• Good enough for climate?
• Validation
• Climate parameters
• Temperature
• Humidity
• Clouds
• Carbon Dioxide
• Carbon Monoxide
• Methane
• Sulpher Dioxide
• Aerosols (dust, ash)
• Cirrus clouds

3
Climate Monitoring with Weather Satellites

• One thermometer, humidity sensor for the globe
• Weather satellites require dense coverage Þ
  expensive sensor with high throughput
• If hyperspectral weather satellites are well
  calibrated, and intercompared will they be
  accurate enough? (Big savings.)
• Problems doing climate with weather satellites
• Calibration very expensive
• Validation what is truth?
• U.S. and Europe now providing different
  hyperspectral sounders in different orbits
• Large data volumes make reprocessing expensive
• Advantages of weather satellites
• Weather satellites appear to be here to stay
• Calibration by sensor vendors is improving
• Technologies are frozen for 15 years at a time
  (part of being expensive)

4

5

6
Climate Model Uncertainties
7
Knowledge of Radiative Forcings
8
AIRS vs Blackbody Radiances

• Black 300K Red Sept. 2003 mean
  radiance, 0 to -10 deg. Lat.
• Green 270K Blue Sept. 2003 mean
  radiance, -40 to -50 deg. Lat.
• Magenta 220K

300K
300K
270K
270K
220K
220K
9
An AIRS Spectrum
10

11

12

13
Radiative Transfer Equation
14
Why AIRS Needs to Last Beyond 2009
15

AIRS on AQUA at TRW
IASI on METOP at Alcatel
16
EOS Aqua was launch on 4 May 2002 into a 705km
altitude sun-synchronous orbit. AIRS IR
calibration started on 13 June 2002 Routine
operations started 1 September 2002 Expected end
of life 2009
17
Atmospheric Spectroscopy Lab Computers
100 cpus, Linux OS Gigabit backbone network 20
Tbytes On-Line 22 Tbytes On Tape (soon) Ingest
33 Gbytes/day On-line database of AIRS clear
FOVS (0.7 Tbytes)
18
AIRS-AMSU-HSB SCIENCE TEAM
Dr. Mous Chahine NASA/JPL Dr. George
Aumann NASA/JPL Dr. Roberto Calheiros
INPE/BSA, Brazil Dr. Alain Chedin CNRS,
France Dr. Catherine Gautier UC Santa
Barbara Dr. Mitch Goldberg NOAA/NESDIS Dr.
Eugenia Kalnay U of Maryland, College
Park Dr. John Le Marshall Bureau of
Meteorology, Australia Dr. Larry McMillin
NOAA/NESDIS Dr. Rolando Rizzi University of
Bologna, Italy Dr. Hank Revercomb
University of Wisconsin Dr. Philip Rosenkrans
MIT Dr. William Smith NASA/LaRC Dr.
David Staelin MIT Dr. Larrabee Strow
U of Maryland, Baltimore County Dr. Joel
Susskind NASA/GSFC
19
Instrument Parameters

• Grating spectrometer
• ?/? ? 1200
• ? 3.7 15.4 microns, ? 650 2700 cm-1
• 13 lines/mm grating, using order 3-11
• Spectrometer cooled to 155K
• HgCdTe detectors (4482 of them) cooled to 58K
• - 49.5 degree ground swath
• FOV is 13.5 km from 705.3 km orbit
• Calibration with space, blackbody views every
  2.67 sec (scan cycle)
• Each 2378 channel spectrum taken in 22.4 msec
• 30 spectra a second!
• 30 Gbytes/day (stored next door)
• 22 Tbytes of data over 2 years (8TB on-line, 0.5
  Tbyte clear)
• 4 visible channels with 2.3km footprint

20

21

22
AIRS Optics
23
AIRS Ray Trace
24
Focal Plane Map
25
An AIRS Spectrum(Color Coded by Focal Plane
Array)
26

27
Ready for Integration on AQUA
28
(No Transcript)
29

30

31
IVAN
32
AIRS Viz Image of Ivan
33
Stability of AIRS Radiometry

• Clear FOVS, ocean only
• Compare observed 2616 cm-1 radiance to
  computed radiance using NOAA RTG-SST product
• RTA-SST tied to bulk SST measured by large
  network of tropical buoys

Slope is 0.008 K/year, so no measurable
radiometric drift Offset due to day/night
aliasing and evaporative cooling of skin This
result is key to using AIRS for climate studies
at other frequencies sensitive to air
temperature, water abundance, and minor gases.
34
Stability of AIRS Frequencies

• Need frequency stability for climate studies as
  well
• AIRS exhibits seasonal frequency cycle,
  peak-to-peak is 0.4 of SRF width!
• Superimposed on seasonal cycle is a smooth drift
  of 0.1-0.3 of SRF width per year.
• Seasonal cycle equiv. to 0.7ppm CO2 (out of 0-6
  ppm variability)
• Long term drift (0.2) equiv. to 0.3 ppm of CO2
  (out of 3 ppm per year)

35
Minor Gas Retrievals with Airs
CO2, CH4, SO2 L. Strow, S. Hannon, S.
DeSouza-Machado, Simon Carn UMBC Physics and
JCET CO Wallace Mcmillan, Juying Warner (UMBC
Physics Department and JCET) and Chris Barnet
(NOAA-NESDIS)
AIRS was not designed to measure minor gases
except for O3

• CO2 Monthly zonal averages, soon will have a
  2-year record (clear,ocean only)
• CH4 Global 2 degree maps zonal averages
  (clear, ocean only)
• SO2 Case studies (about 8 eruptions studied so
  far), no monitoring
• CO Large scale retrievals using C. Barnets
  AIRS science code in addition to monthly
  averages. Land and ocean, uses cloud-cleared
  radiances.

36
CO2 Detection with AIRS

• The AIRS dataset used here is a clear-only, ocean
  set of FOVs that are subsetted at UMBC from the
  L1b data.
• These data are averaged on a 2 by 2 degree
  lat/long grid.
• Most work done with the 792 cm-1 channel, some
  with the 2390 cm-1 channel.
• The minor gas amounts are computed by shifting
  the CO2 profile until the brightness temperature
  bias is zero for the particular channel sensitive
  to CO2.
• Use ECMWF T(z) and H2O(z) (adjusted for observed
  total column water)
• Fit for effective SST
• Kernel function peaks around 500 mbar, little
  sensitivity in boundary layer (why you need OCO).
• We have also derived mean monthly CO and CH4
  amounts with this same technique.
• Most of the results here use zonal averages
  derived from a 2 x 2 degree monthly average
  lat/lon grid.

37
AIRS Jacobians for Minor Gases
CO_2 measurement done with 792 cm-1
channel. Planck T-dependence almost cancels
Boltzmann T-dependence
38
Aug. 2002 April 2003 CO2 Monthly Zonal
AveragesUsing the 792 cm-1 Channel
39
Zonally Averaged CO2 Variability with Time
40
Fit of 20 months to sinusoidal function(secular
trend previously determined)

• CO2 a bsin(2?t/12 c)
• t time in months

41
Zonally Averaged CO2 Variability with Time
42
20 Months of Zonally Averaged CO2791.75 cm-1
Channel
43
20 Months of Zonally Averaged CO22390.11 cm-1
Channel
44
AIRS Global Mean CO2Weighted by cos(LAT)
Estimated bias is 5ppm ( 0.17K) based on
NOAA-CMDL global mean for 2003.
0.1K!
45
Comparison of CO2 Using Different Channels(2390
cm-1 bias was scaled to equal mean value of 792
cm-1 bias)
46
NOAA CMDL vs AIRS for Dec. 2002(CMDL has not
made these data available for 2003 as yet.)
47
VERY Preliminary CO2 Map for May 2003
Obs ppm - 370
48
CO2 Future Work

• Extend to land
• Improve S/N with more channels (better spectral
  cal of 4.3 micron channels will make this
  possible)
• Can T(z) and CO2 be retrieved simultaneously w/o
  ECMWF?
• Monthly averaging probably hides traces of
  source/sinks?
• IR radiances contain lots of interesting climate
  signatures!

49
CO Retrievals from AIRS Radiances

• CO retrieval part of standard T(p), Q(p)
  retrieval, implemented in Chris Barnets science
  code running on our Linux cluster at UMBC.
• Retrieves CO column that nominally peaks 500
  mbar, very little boundary layer sensitivity.
• Sept. 2002 global retrievals
• INTEX support real-time retrievals of CO to
  aid aircraft flight-lines during INTEX. Data
  analyzed less than 24 hours after being recorded.
  (Thanks to Mitch Goldberg and Walter Wolf for
  providing AIRS data from their real-time feed.)
• AIRS NOT optimized for CO only a few CO lines
  are recorded.
• Validation with in-situ sensors and MOPITT
  suggest 15 column accuracies.

50
CO in AIRS Spectra
51
AIRS CO Kernel Functions

• 500

52
AIRS
MOPITT
53
UMBC AIRS Daily Global CO

• 80 of the planet seen each day, peak
  sensitivity 500mb
• preliminary validation says good to 15 at
  500mb

54
UMBC AIRS CO during INTEX
AIRS
fires
Asian pollution
USA pollution
MOPITT

• Standard retrievals on AMSU footprint, binned to
  a 1 degree grid
• preliminary validation says good to 15 at 500mb

55
Methane from AIRS

• 19 months of methane column in the 200-700 mbar
  range
• ECMWF Used for atmospheric temperature and
  humidity
• Monthly averages over ocean using clear FOVs
• Scale is in change in methane from reference
  value
• ALL RESULTS SHOWN HERE ARE PRELIMINARY!

56
Zonal Averages of Methane19-month record (Sept.
2002 March 2004) Constant biased to North.
Hemis. winter
? CH4 a bsin(2pi? t/12 c), ? t month
index
Fit assumed no secular trend!
57
09/2002 vs 09/2003
58
11/2002 vs 11/2003
59
01/2003 vs 01/2004
60
03/2003 vs 03/2004
61
(No Transcript)
62
(No Transcript)
63
(No Transcript)
64
(No Transcript)
65
(No Transcript)
66
(No Transcript)
67
(No Transcript)
68
(No Transcript)
69
(No Transcript)
70
(No Transcript)
71
(No Transcript)
72
(No Transcript)
73
(No Transcript)
74
(No Transcript)
75
(No Transcript)
76
(No Transcript)
77
(No Transcript)
78
(No Transcript)
79
(No Transcript)
80
Separation of Dust from Cirrus
Imaginary Index of Refraction of Ice and Dust

• Both ice and silicate absorption small in 1200
  cm-1 window
• In the 800-1000 cm-1 atmospheric window
• Silicate index increases
• Ice index decreases
• with wavenumber

Volz, F.E. Infrared optical constant of
ammonium sulphate, Sahara Dust, volcanic pumice
and flash, Appl Opt 12 564-658 (1973)
81
Silicate (ash cloud) Detection at Anatahan
Image is ECMWF bias difference of 1227 cm-1 984
cm-1
Note slope
82
Cirrus Detection at Anatahan
Image is bias ECMWF bias difference of 1227 cm-1
781 cm-1
83
SO2 Not in Standard AIRS-RTA
SO2
Volcanic Ash
Red Residuals after fitting ash cloud parameters
and SO2 column Green Residuals for clear sky
calculation, no SO2
84
Mt. Etna Eruption Ash and SO2
85
Anatahan Eruption Differential Movement of Ash
vs SO2 Cloud
Ash cloud
SO2 cloud
86
Dust Detection 1231 and 966 cm-1
87
Optical Depth of Dust Longwave Only
MODIS Visible
88
Optical Depth of Dust Shortwave Only
MODIS Visible
89
Spectral Fitting Bias and Standard Dev.
SARTA-Scattering used to fit optical depths,
particle size. Results with/without floating
dust altitude (CO2 slicing) and SST
similar Log-normal distri-bution, particle size
spread 1 micron Used ECMWF fields, not
retrievals, for T(z), Q(z), ozone, etc. Only
longwave window channels are used to fit for dust
para-meters
90
Sept. 2002
91
Oct. 2002
92
Nov. 2002
93
Dec. 2002
94
Jan. 2003
95
Feb. 2003
96
Mar. 2003
97
Apr. 2003
98
May 2003
99
June 2003
100
July 2003
101
Aug. 2003
102
Expanded Scale Shows Biases of -3K!
103
Standard Deviation of Dust Bias
104
Cirrus Retrievals SARTA-Scattering RTA
105
Mean bias for various cirrus particle habits
O C in K
Wavenumber (cm-1)
106
Caption of Previous Figure