Line-by-line model development in support of the JCSDA PI – Eli Mlawer (AER) Co-PI – Vivienne Payne (JPL) AER contributors – Matt Alvarado, Karen Cady-Pereira,

Line-by-line model development in support of the JCSDA

PI – Eli Mlawer (AER)

Co-PI – Vivienne Payne (JPL)

AER contributors – Matt Alvarado, Karen Cady-Pereira, Jean-Luc Moncet, Gennady

Uymin, Dan Gombos, and Alan Lipton

Joint Center for Satellite Data Assimilation Workshop

June 5, 2013

Copyright 2013, Supported by JCSDA external research program through NASA contract NNH11CD78C – “Maintaining High Quality Spectroscopy for the Community Radiative Transfer Model”

• Satellite data assimilation depends on accurate IR spectroscopy– Reducing uncertainties in spectroscopic line parameters and continua is critical to

improving the use of satellite data in numerical weather prediction (NWP) and climate models.

• JCSDA Community Radiative Transfer Model (CRTM) is trained using AER line-by-line models

– Both OPTRAN-based CRTM and new version developed with AER’s OSS approach

• Two LBL models – LBLRTM and MonoRTM (Clough et al., 2005) Line-by-Line Radiative Transfer Model (LBLRTM)

• For general applications, consistent physics for all spectral regions from MW to UV

• Reference standard for model intercomparisons in the thermal IR (e.g., CCMVal, CIRC)

• Basis of retrieval algorithms for IASI, TES, ...

Monochromatic Radiative Transfer Model (MonoRTM)• For applications requiring higher accuracy (e.g. narrow channels such as in microwave)

Spectroscopic Input to AER’s LBL Models

• AER’s line parameter database - Built from HITRAN with selected upgrades• MT_CKD continuum model

– Includes water vapor continuum model derived from field and lab measurements; used by most RT codes

JCSDA and Line-by-Line Modeling at AER

• Release of LBLRTM_v12.2

– Upgrades to CO2, O2, spectroscopy

– Over 300 downloads from AER RT Model webpage (rtweb.aer.com)

• Validation studies

– Recent spectroscopic updates to LBLRTM using IASI measurements • Alvarado et al., ACPD, 2013

– Spectroscopy related to mid/upper tropospheric water vapor• Results from high-altitude (5300 m) field campaign in Chile - RHUBC

• Ongoing project – upgrade of features in MonoRTM

– Broadening of O2 lines by H2O

Recent Developments in AER’s LBL Modeling

Recent Updates to LBLRTM

Spectroscopy

•CO2 -First order line coupling parameter (P-,Q-, and R-branches)

computed using database of Lamouroux et al. (2010)-Lines between 5970 cm-1 and 6400 cm-1 were updated based on

the work of Devi et al. (2007a, 2007b)

•O2

-Adopted recent HITRAN linelist update (replacement of HITRAN 2008 list)

-Quadrupole linelist of Gordon et al. (2010) was adopted (7470-8200 cm-1)

Model release – LBLRTM_v12.2•Released in November 2012

-Spectroscopy updates, increased use of structures, bug fixes, ...-Widely downloaded by community

Improvement due to new O2 linelist

•Evaluated using spectra measured by Bruker FTS in Lamont, OK

- Part of the Total Carbon Column Observing Network (TCCON)- Measures from 3900-15500 cm-1 with 0.02 cm-1 resolution- Analysis included 75 cases from various seasons, zenith angles, aerosol and H2O loadings

Wavenumber (cm-1)

December 20, 2009, SZA = 70°

Improvement due to new O2 linelist

Wavenumber (cm-1)

HITRAN 2008 lines New Linelist

December 20, 2009, SZA = 70°

Recent Updates to LBLRTM

Spectroscopy

•CO2 -First order line coupling parameter (P-,Q-, and R-branches)

computed using positions and intensities in Lamouroux et al. (2010)

-Lines between 5970 cm-1 and 6400 cm-1 were updated based on the work of Devi et al. (2007a, 2007b)

•O2

-Adopted recent HITRAN linelist update (replacement of HITRAN 2008 list)

-Quadrupole linelist of Gordon et al. (2010) was adopted (7470-8200 cm-1)

Model release – LBLRTM_v12.2•Released in November 2012

-Spectroscopy updates, increased use of structures, bug fixes, ...-Widely downloaded by community

AER RT Model Downloads for 5/20-5/31

Includes downloads from NIST, Harvard, UMBC, Caltech, U. Idaho, 2 U.S. private companies, Canada, Austria, Pakistan, Belgium

• Rigorous validation of recent spectroscopic updates to LBLRTM against a global dataset of 120 near-nadir measurements from IASI.

• The performance of LBLRTM v12.1 is compared to a previous version (LBLRTM v9.4+) to test the impact of the latest updates to the line parameters and the CO2 and H2O continua (including the addition of P- and R-branch line coupling for CO2)

- Alvarado et al., ACPD (2013)

Validation Study of LBLRTM vs. IASI

IASI Closure Study Methodology

Schematic of the retrieval procedure. The dashed arrows show additional retrievals performed to assess the

consistency of CO2 in the IASI spectral range.

120 clear-sky, nighttime, over ocean IASI profiles (a subset of Matricardi, 2009)

- minimize potential errors from clouds, surface emissivity, and NLTE effects.

Systematic spectral residuals after retrievals indicate errors in the spectroscopy.A priori profiles:

•Temperature: ECMWF adjusted between 10 mbar and 0.1 mbar (Masiello et al., 2011).

•H2O: ECMWF model.•O3: ECMWF model scaled by OMI.•CO2, N2O, CH4, and CO: Aura TES

climatology

The addition of P- and R-branch line coupling improved performance in the ν2 band of CO2

IASI Scan #754

Mean ResidualsLBLRTM v12.1

Mean ResidualsLBLRTM v9.4+

The spectroscopy updates alter the temperature profiles retrieved using the ν2 band

Right: Mean and std. dev. of the

differences between the temperature

profiles retrieved with LBLRTM v12.1 and

v9.4+. Only the cases that converged for all

four model/band combinations are

included.

Updates to the MT_CKD continuum have improved performance at the ν3 bandhead

IASI Scan #754

Mean ResidualsLBLRTM v12.1

Mean ResidualsLBLRTM v9.4+

Note that green ν3 region was not used in temperature retrievals here

The ν2 and ν3 temperature retrievals in LBLRTM v12.1 are remarkably consistent.

Mean and std. dev. of the differences between

the retrieved temperature profiles.

Right panel: the ν2 retrieval was smoothed

with the ν3 averaging kernel and

retrieval(Rodgers and Connor, 2003).

€

ˆ x smooth ν 2= ˆ x ν 3

+ Aν 3ˆ x ν 2

− ˆ x ν 3( )

Residuals in the H2O ν2 band are improved, especially in P-branch

IASI Scan #754

Mean ResidualsLBLRTM v12.1

Mean ResidualsLBLRTM v9.4+

Residuals in the H2O ν2 band are improved, especially in P-branch

Updated H2O spectroscopy in LBLRTM v12.1 impacts retrieved H2O profiles

Right: Mean and std. dev. of the ratio between the H2O

profiles retrieved with LBLRTM v12.1 and v9.4+. Only the 122

cases that converged for temperature and

H2O in both models are included.

Residuals in the 616 JCSDA assimilated IASI channels are substantially improved

IASI Scan #754

Mean ResidualsLBLRTM v12.1

Mean ResidualsLBLRTM v9.4+

Conclusions from Alvarado et al.

• The improved CO2 spectroscopy in LBLRTM v12.1 can alter the retrieved temperature profiles by 0.5 K or more.

• The LBLRTM v12.1 CO2 spectroscopy is remarkably consistent between the CO2 ν2 and ν3 bands.

– Systematic residuals remain in the ν2 Q-branches and at the ν3

bandhead.

• The H2O spectroscopy is improved in both the P- and R-branches of the ν2 band in LBLRTM v12.1, but significant systematic residuals remain.

– Using a more realistic, vertically-varying HDO profile may reduce the P-branch mean residuals for scans with high water vapor.

Existing satellite- and ground-based observations can validate most, but not all, infrared spectroscopy of interest.

Spectral Cooling Rate Profiles in the Infrared for MLS

Far-Infrared

Infrared Transmittance

Scientific Motivation for Radiative Heating in Underexplored Bands Campaigns (RHUBC)

• Radiative heating/cooling in the mid-troposphere modulates the vertical motions of the atmosphere– This heating/cooling occurs primarily in water vapor absorption

bands that are opaque at the surface - essentially unvalidated

• Approximately 40% of the OLR comes from the far-IR– Until recently, the observational tools were not available to

evaluate the accuracy of the far-IR radiative transfer models

• Upper troposphere radiative processes are critical in understanding the radiative balance of the tropical tropopause layer and the transport of air into the stratosphere

• These processes need to be parameterized accurately in climate simulations (GCMs)

RHUBC CampaignsMain objective: Use radiative closure to reduce uncertainties in H2O spectroscopy

(H2O continuum absorption model, line parameters (e.g. strengths, widths))

RHUBC-I • ARM North Slope of Alaska (Barrow)• February - March 2007• ~80 radiosondes launched• 2 far-IR/IR interferometers• Microwave radiometers for PWV

RHUBC-II• Cerro Toco (~5350 m), Atacama, Chile• August - October 2009• 3 far-IR / IR interferometers• ~100 radiosondes, extremely low PWV• 1 microwave radiometer for PWV• 1 sub-millimeter FTS

RHUBC Precipitable Water Vapor (PWV) Values

For reference, PWV for US

Standard atmosphere is

14.3 mm.

Transmission in the Infrared

Uncertainty in the WV Continuum in Far-IR

1st PrinciplesSHEBA

RHUBC-I

Far-infrared Spectroscopy of the Troposphere (FIRST)• PI - Marty Mlynczak, NASA-LaRC• Michelson interferometer• 100 - 1600 cm-1 (resolution ~0.64 cm-1)

Radiation Explorer in the Far Infrared (REFIR)• Italian collaboration (RHUBC lead - Luca Palchetti)• Fourier Transform Spectrometer• 100 - 1500 cm-1 (resolution ~0.50 cm-1)

Smithsonian Astrophysical Observatory FTS• PI - Scott Paine• 300 GHz – 3.5 THz (3 GHz resolution)

U. Cologne HATPRO• 7 channels from 22.2 -31 GHz, 7 channels from 51-58 GHz

Atmospheric Emitted Radiance Interferometer (AERI)• ARM Instrument developed at U. Wisconsin• 550 - 3000 cm-1 (resolution ~0.5 cm-1)

RHUBC-II Instruments

Spectral Observations 170 GHz (5.6 cm-1) to 3 µm (3000 cm-1)

SAO FTS (Smithsonian)FIRST (NASA/LaRC)

REFIR (Italy)AERI (UW)

First ever measurement of the entire infrared spectrum from 3 to 1780 μm!

Improving Sonde H2O Profiles• Sonde H2O values have known accuracy issues

• Other measurements can be used to improve sonde profiles

- Vaisala RS92 H2O correction derived using e.g. frost-point hygrometer Miloshevich et al. (2009) - may not apply to RHUBC sondes

Sonde too

moist

Sonde too dry

RHUBC-II PWV

Improving Sonde H2O Profiles• Sonde H2O values have known accuracy issues

• Other measurements can be used to improve sonde profiles- Vaisala RS92 H2O correction derived using e.g. frost-point hygrometer Miloshevich

et al. (2009) - may not apply to RHUBC sondes

• Analysis from ARM SGP: IR radiative closure improved after sonde H2O profile scaled to agree with PWV retrieved from MW radiometer (22 GHz line)

- 22 GHz line too weak to provide information for RHUBC

• Spectroscopy of 183 GHz line has low uncertainty

– Clough et al. (1973), Payne et al. (2008, 2011)

GVRP: channels centered at 170, 171, …, 183, 183.3 GHz

Optical Depths for PWV ~0.3 cm

• Approach: Retrieve water vapor profiles using GVRP measurements- Optimal estimation approach

- Miloshevich et al. (2009) corrected sonde used as initial guess / prior

Water Vapor Profile Retrieval from GVRP

Impact of WV Profiles on Far-IR Radiances

RHUBC-II: Comparison with SAO-FTS (sub-mm)

figure: Scott Paine

RHUBC-II: Comparison with HATPRO (μwave)

all RHUBC sondesfigure: Gerrit Maschwitz

Impact on downwelling brightness temperature of adding O2 isotopologue lines to default linelist

Impact on upwelling (US Std)

MonoRTM upgrade

• Currently ongoing – release in July

• Objectives

- Support OSS-CRTM development by making MonoRTM more modular

- directly calls more LBLRTM subroutines

- utilizes same binary line parameter file as LBLRTM

- Add capability to utilize line broadening coefficients due to collisions from a specific molecule (when available)

• e.g. broadening of: CO2 due to H2O; O2 due to H2O in microwave

- Implement speed dependent line shape

- Augment default microwave linelist (as needed)

- Various bug fixes

H2O broadening of O2 in microwave

• New measurements by Brian Drouin of JPL

Paper submitted - Drouin, Payne, Oyafuso, Sung, Mlawer: “Pressure broadening of oxygen by water”

Change in brightness temperature from including

this effect for atmosphere with PWV = 3 cm

MonoRTM upgrade

• Currently ongoing – release in July

• Objectives

- Support OSS-CRTM development by making MonoRTM more modular

- directly calls more LBLRTM subroutines

- utilizes same binary line parameter file as LBLRTM

- Add capability to utilize line broadening coefficients due to collisions from a specific molecule (when available)

• e.g. broadening of: CO2 due to H2O; O2 due to H2O in microwave

- Implement speed dependent line shape

- Augment default microwave linelist (as needed)

- Various bug fixes

Future work – evaluate temperature dependence of O2 line widths

• Comparison of HATPRO measurements from site in Greenland with MonoRTM calculation

• Shows possible line width temperature dependence issue at 52.3 GHz

- from Miller et al. (2013)

Acknowledgements

• S. A. Clough, Clough Radiation Associates• Marco Matricardi, ECMWF• Scott Paine, Smithsonian Astrophysical Observatory• Joint Center for Satellite Data Assimilation (JCSDA)• NASA• Gerrit Maschwitz• Dave Turner, NOAA NSSL

AERI-ER

ARM Instruments for RHUBC-II

GVRP

MPL

MFRSR

Vaisala Ceilometer

RadiometersSondes

Met station

Far-IR Analysis: Results

• 17 cases used in study; H2O from sondes scaled by GVR +/- 7 PWV retrieval• Adjustments made to water vapor continuum and selected line widths• Delamere et al., JGR, 2010

Average AERI Radiances

AERI-LBLRTM Residuals Before RHUBC-I

Residuals After RHUBC-I

Site location

RHUBC-II, Cerro Toco, Chile