Q-6 HS Sounder Constituent Profiling Capabilities Cirrus Cloud

Q-6: HS Sounder Constituent Profiling Capabilities Cirrus Cloud Boundaries from the Moisture Profile W. Smith 1, 2, B. Pierce 3, and Z. Chen 2 1 University of Wisconsin, 2 Hampton University, 3 NOAA/CIMSS Satellite Hyperspectral Sensor Workshop Rosentiel School of Marine & Atmospheric Science (March 29 -31, 2011

Content of Presentation • Review Hyperspectral sounding capability for Green House Gases (GHG) measurements. • Present a physical/statistical technique for GHG sounding utilizing global chemistry model (RAQMS) profiles and the RTTOVS Radiative Transfer Model to prescribe GHG-profile/Radiance-spectrum covariance • Demonstrate the capability of current satellite Hyperspectral sounders (e. g. , IASI) profile T, H 2 O, O 3, CH 4, CO, N 2 O. • Discuss the need for Geo-Hyperspectral sounders to observe boundary layer GHG concentrations and tropospheric gas fluxes.

Sensitivity of IASI to GHGs Tskin – Tair = 0 K Tskin – Tair = 15 K IASI Water Vapor Tskin – Tair = 0 K Tskin – Tair = 15 K IASI Carbon Dioxide Surface Skin / Surface Air Temperature Contrast Increases Satellite Sensitivity in Lower Atmosphere

Question #6 Atmospheric Constituent Profiles From IR Radiance Spectra Motivation of this Retrieval Approach • Use statistical covariance between gas mixing ratio profiles and the radiance spectrum to obtain GHG gas profiles • Procedure adds statistical information as needed for profile retrieval through the cross-gas and crosstemperature correlations contained in the radiance spectrum • Statistical information provided by a global scale model with interactive chemistry/weather dynamics physics and radiative transfer model.

High-performance Instrumented Airborne Platform for Environmental Research (HIAPER), Pole-to-Pole Observation (HIPPO) III PI: Steven C. Wofsy, Harvard University NCAR G-V aircraft March 20 -April 20, 2010 National Science Foundation (NSF)-sponsored effort to study the distribution of greenhouse gases and black carbon in the atmosphere. High-accuracy measurements of greenhouse gases and black carbon particles from the top of the troposphere to the earth's surface and pole-to-pole. http: //www. eol. ucar. edu/deployment/field-deployments/field-projects/hippo_global_3

Developed by Brad Pierce (NOAA/CIMSS) Denver to Anchorage 03/24/2010 RAQMS O 3 curtain with HIPPO 3 in-situ O 3 (Spackman, NOAA/ESRL)

IASI Granules Overlaying HIPPO Flight Tracks

GHG Profile Retrieval Algorithm qret = qo + (rm-ro)C C = (R’TR’ + λETE)-1 R’TQ’ q = atmospheric profiles (Temp. H 2 O, O 3, CH 4, CO, N 2 O) C = global ensemble statistical covariance about the mean profile qo R = global ensemble of Principal Component (PC) scores (deviation from their ensemble mean values) of the radiances calculated from atmospheric profiles, q. Q & R = global ensemble of T/GHG profiles & calculated radiance spectra PC scores. ( )’ = deviation from the ensemble mean values, qo and ro ETE = statistical covariance of spectral radiance PC score noise λ = matrix conditioner assumed to be 0. 05 % of the variance of the error for the first (i. e. , most important) PC score. Spectral Range: 650 -2250 cm-1 & Spectral Spacing: 0. 25 cm-1 = 6401 Channels 300 Radiance Principal Components Used for Retrieval

Statistical BIAS Correction When using global ensemble statistics, the retrieved profile will have significant bias towards the global mean for scale vertical structure modes not resolved by the radiance observations (i. e. , modes within the null space ). This bias, or null space error, can be minimized by producing a retrieval from a error-free radiance spectrum simulated from a known atmospheric condition, which is more representative of the atmospheric condition being retrieved than is the mean profile. In practice, a numerical prediction model background field can be used to provide a more representative, than the global mean, known reference atmospheric condition. Thus, an estimate of the “null-space” error correction is given by: Cor= (qknown – qret/known) where in this case the RAQMS profile is used to represent a known profile and rknown is the vector of PC scores corresponding to the known noise free radiance calculated from the known profile

Cloud Height Estimation Retrieved values Set Equal to Missing Below Cloud Level ( Tiasi < Traqms for p>pcld )

Example Profiles

Corrected Vs Uncorrected

Observed Vs Simulated (Q)

Total Column Concentration IASI Vs RAQM (U) Gas H 2 O IASI – RAQMS Difference (% std) Mean Std - 13 28 O 3 5 22 CO - 26 28 CH 4 - 20 23 N 2 O -8 20

IASI Vs Aircraft Measurements (CO)

IASI Vs Aircraft Measurements (O 3)

Geo-HS Is What is Really Needed!

Conclusion • A retrieval technique has been developed to produce vertical profiles of GHG using the statistical covariance between the GHGs and the radiance spectrum • The technique was applied here by applying the statistical structure of GHG profiles and simulated IASI radiances provided by the global scale RAQMS for the HIPPO experiment where aircraft ground truth is available • RAQMS predicted GHG profiles were successfully validated by comparison of the profiles with retrievals from IASI observations obtained during the HIPPO experiment • Finally, a mesoscale WARF model Hurricane Ike simulation was used to illustrate how the four dimensional water vapor distribution can be observed by a Geostationary Hyperspectral Imaging Spectrometer I dedicate this work to my colleague for more than 40 years, Moustafa Chahine. Mous was a tenacious remote sensing scientist/experimenter/leader; as a competitor he made my professional life much richer than it would have been without him. Thank you Mous, I will miss you. May you keep active in your new life.