Temperature and Pressure Measurements from ACEMAESTRO

1
Temperature and Pressure Measurements from
ACE-MAESTRO

• Caroline Nowlan
• Harvard-Smithsonian Center for Astrophysics
• University of Toronto
• Tom McElroy, Environment Canada
• Jim Drummond, University of Toronto
• EPS 238
• May 1, 2008

2
Contents

• The instrument
• Radiative transfer modeling
• Spectral fitting
• Some data

3
Atmospheric Chemistry Experiment

• ACE is on SCISAT satellite (650 km
  circular, 74 orbit)
• Launched August 12, 2003
• Two primary instruments on-board
• ACE-FTS (Fourier Transform Spectrometer)
• ACE-MAESTRO

• Goal To investigate the chemical and dynamical
  processes that control the distribution of ozone
  in the stratosphere and upper troposphere with a
  particular focus on the Arctic winter
  stratosphere. (Bernath et al., 2005)

4
Occultation Distribution
5
MAESTRO

• Measurement of Aerosol Extinction in the
  Stratosphere and Troposphere Retrieved by
  Occultation

6
MAESTRO Spectrometers
7
Slit Function and Field-of-View
8
Beer-Lambert Law andOptical Depth
Length L
I0
I

• t optical depth
• absorption cross-section (cm2/molecule)
• n density (molecules/cm3)
• L length of path (cm)

9
Transmission for a tangent height of 15 km
10
Methodology

• O2 is well-mixed in atmosphere
• MAESTRO measures three bands of O2
• A band at 762 nm
• B band at 690 nm
• g band at 632 nm

11
O2 Bands in Visible-NIR
A-band
1
0.5
0
760
770
780
1
0.5
0
760
770
780
1
0.5
0
760
770
780
12
Remote Sounding Inverse Problem

• We have y (spectral data)
• We want x (density profile)
• Requirements for inversion
• Forward model (F)
• Retrieval algorithm

13
Global Fit Retrievals

• Simultaneous fitting of every spectrum in an
  occultation
• Arrange all spectra into one giant vector
• Nonlinear iterative retrieval
• Dx (KT Se-1 K Sa-1 g H)-1 (KTSe-1 Dt
  Sa-1(x xa) - g H x)
• x retrieved parameter
• Se measurement error covariance matrix
• K dt/dx
• H smoothing matrix
• g smoothing weighting
• xa a priori profile (forecast/reanalysis
  profile)
• Sa a priori error covariance matrix

14
O2 Modelling

• Must model at high-resolution because of
  nonlinear absorption (saturated lines), and
  across entire band (300 individual lines)
• Problem We want to calculate analytic
  differentials

15
Fast Line-by-Line

• Separate from retrieval
• Run line-by-line code for 15 pressures and 5
  temperatures and construct cross-section table
  with elements
• l p T
• During retrieval, at each iteration
• Use 3-D cubic interpolation to calculate
  high-resolution cross-sections for
  pressure-temperature combinations on 1 km grid

16
Correlated-k Approximation
17
Model Errors
FLBL
Correlated-k
18
Error Analysis

• Error sources
• Model parameters
• O2 spectroscopy
• Slit function
• Pointing
• Forward model error (approximations)
• Noise

19
Raw Data VIS spectrometer
20
Optical Depth Spectra VIS
21
Optical Depth Spectral Fitting
22
(No Transcript)
23
(No Transcript)
24
Density Averaging Kernels
25
In Conclusion

• Correlated-k is a high-accuracy fast way to model
  O2
• Global fitting retrieval algorithm required
• Error analysis estimates1 pressure and 2 K
  temperature error
• Satellite data comparisons show
• Biases
• 2 5 in pressure, lt 5 K in temperature
• RMS differences
• 2 10 in pressure, 5 10 K in temperature
• B band has been very useful ? used entirely below
  30 km