Deriving the Instrument Transfer Function from OMI Solar

Deriving the Instrument Transfer Function from OMI Solar Observations and Its Implications for Ozone Retrievals Kang Sun, Xiong Liu, Zhaonan Cai, Guanyu Huang, Gonzalo González Abad, and Kelly Chance Harvard-Smithsonian Center for Astrophysics Kai Yang, UMD GSICS Annual Meeting, Mar. 22, 2017 1

Instrument transfer function (slit function) Nadir OMI detector array, UV 1 band Native resolution spectrum: Observation by OMI: 2

Why is ITF a challenge? • OMI slit functions were thoroughly measured preflight (Dirksen et al. 2006) os Cr a sp k( rac s-t • Is the on-orbit slit function the same to preflight measurements? l) tia e dir on cti on i t c e r h di t g n e avel W • Is the slit function stable on-orbit? • Is it possible to derive onorbit slit function and improve retrieval? 3

Derive ITF by solar observations • The slit function shape retrieved by fitting OMI solar spectra with a highresolution solar reference spectrum Slit function Inverse Forward High resolution reference solar spectrum Measured solar spectra 4

Fitting on-orbit slit function • Gaussian function: • Super Gaussian function: • Preflight ITF with homogeneous stretch: k = 2, Changing w w = constant, Changing k Stretch • 1. spectrally averaged Beirle et al. AMT 2017 • 2. spectrally resolved

On-orbit temporal variation: fitting windows • Divide each OMI band into four windows • Assume standard Gaussian slit function for each window 6

Fitted slit function width On-orbit temporal variation: cross track dependence 7

On-orbit temporal variation: cross track dependence 8

Solar function change or solar cycle? De. Land Marchenko, http: //sbuv 2. gsfc. nasa. gov/solar/omi/ 9

On-orbit temporal variation: solar cycle and RA Non-RA rows 10

On-orbit temporal variation: solar cycle and RA RA rows 11

On-orbit slit functions differ from preflight! Cross-track (1 -30) Cross-track (1 -60) 12

SAO ozone profile retrievals Stretched preflight 2 • Test SAO ozone profile retrievals using • Gaussian, operational • Super Gaussian • Preflight slit function • Stretched preflight (spectrally averaged) • Stretched preflight (spectrally resolved) Stretched preflight 1 13

Ozone profile validations – cross-track pattern Operational SAO ozone profile data Many other products! • Note: each column is the medium bias of 50 – 200 pairs of collocated OMIsonde profiles 14

Summary • Derived OMI on-orbit slit functions complicated by solar activities at certain wavelengths • Slit functions of non-RA rows stable over time; RA had impact on slit functions for < 300 nm • On-orbit slit function widths differ from preflight, varying by cross-track positions • Derived slit functions show better cross-track consistency in ozone retrieval 15

Backup: validation using ozone sondes (2004 – 2008) • Cloud fraction < 0. 3 • Dt < 6 hr • Dlat, Dlon <1 • SZA < 75˚ • RMS < 1. 5 • Use averaging kernels 16

Ozone profile validations 17

Stratospheric ozone columns validation 10

Tropospheric ozone columns validation 10

Ozone profile validations – UV 2 only Same issue in UV 2 A single “stretch” of preflight • Note: each column is the medium bias of 10 – 120 pairs of collocated OMIsonde profiles 20

UV 2 UV 1 UV 2 21

22