OMI cloud optical depth contributes to the observed
OMI cloud optical depth contributes to the observed positive bias in surface UV Anders V. Lindfors, T. Mielonen, M. R. A. Pitkänen, A. Arola, J. Tamminen Finnish Meteorological Institute OMI Science Team 2014, Anders Lindfors / FMI
What is known about OMUVB performance? • OMUVB is known to overestimate the surface UV • Discussion has concentrated on aerosols as the reason for overestimation • Sodankylä cloud-free r. MB = 0. 08 Mikko Pitkänen (MSc, 2013) ² comparison in Jokioinen and Sodankylä, matching the overpass time ² cloud classification using sunshine duration, cloud amount, surface solar radiation ² OMUVB performance depends on clouds ² overcast conditions: stronger overestimation ² similar results also in other studies: Weihs et al. (ACP, 2008) Sodankylä overcast r. MB = 0. 29 OMI Science Team 2014, Anders Lindfors / FMI
OMUVB under overcast clouds? • Interest in understanding why there is a systematic, cloud-related overestimation in OMUVB • No proper validation of OMI cloud optical depth (COD) has been done • COD is a primary input to OMUVB calculations • Idea: to compare OMI COD (Aura) with MODIS COD (Aqua) • Aim: to understand more about why OMUVB overestimates in overcast conditions http: //en. wikipedia. org/wiki/A-train_(satellite_constellation) OMI Science Team 2014, Anders Lindfors / FMI
Matching OMI and MODIS CODs OMI 24 x 13 km (nadir) selected footprint in white • • MODIS zoom-in: same area 16 min before selected OMI pixel in white 200— 400 MODIS pixels
OMI cloud optical depth how compare with MODIS? • how to compare CODs from two different instruments? • MODIS 1 x 1 km • OMI 13 x 24 km • exponential relation R vs COD • logarithmic average of COD has been found to be useful • MODIS R 1 R 2 COD 1 CMF 1 COD 2 CMF 2 from MODIS cmp/w OMI COD Figure from Zinner and Mayer (JGR, 2006)
OMI cloud optical depth how compare with MODIS? • how to compare CODs from two different instruments? • MODIS 1 x 1 km • OMI 13 x 24 km • exponential relation R vs COD • logarithmic average of COD has been found to be useful • OMI R 1, 2 COD 1 COD 2 CMF 1, 2 from MODIS cmp/w OMI COD Figure from Zinner and Mayer (JGR, 2006)
CMF = Cloud Modification Factor • CMF = Fall-sky / Fcloudfree • CMF can be averaged (assuming independent pixel radiative transfer): COD 1 ² CMF 1, 2 = (CMF 1 + CMF 2)/2 ² CMFMODIS = CMF 1, 2, …, N CMF 1, 2 = ( COD 2 CMF 1 + CMF 2 ) / 2 ² CMFMODIS cmp/w CMFOMI • radiative transfer model used to calculate CMFMODIS and CMFOMI Science Team 2014, Anders Lindfors / FMI
OMI vs. MODIS (#1): nr of colocated pixels • 10 days: 10— 19 July 2006 • 1. 4 x 106 colocated pixels in total • Only OMI footprints fully cloudy as seen by MODIS were included • Finland is sunny ! OMI Science Team 2014, Anders Lindfors / FMI
OMI vs. MODIS (#2): COD vs. exponent of log-averaged COD • All cases included • 1. 4 x 106 colocations • good agreement • OMI somewhat lower than MODIS for COD>10 OMI Science Team 2014, Anders Lindfors / FMI
OMI vs. MODIS (#3): COD vs. exponent of log-averaged COD • MODIS ice clouds • 500 x 103 colocations • OMI COD somewhat higher than MODIS OMI Science Team 2014, Anders Lindfors / FMI
OMI vs. MODIS (#4): COD vs. exponent of log-averaged COD • MODIS water clouds • 450 x 103 colocations • OMI COD clearly lower than MODIS OMI Science Team 2014, Anders Lindfors / FMI
Undestanding difference between ice and water clouds OMI More backscatter for same optical depth ICE WATE R • OMI cloud model always assumes water clouds • Scattering phase function of ice: more backscatter Ø OMI sees ice clouds as thicker! Ø This explains relative difference between water / ice cloud performance OMI Science Team 2014, Anders Lindfors / FMI
OMI vs. MODIS (#5): CMF vs. latitude • All cloud types • 10 th/90 th percentile limits: COD 1— 80 • OMI CMF higher or at same level as MODIS • Finnish latitudes (60 N): ² small CMF difference of 0. 02— 0. 03 OMI Science Team 2014, Anders Lindfors / FMI
OMI vs. MODIS (#6): CMF vs. latitude • Ice clouds • 10 th/90 th percentile limits: COD 1— 80 • OMI CMF lower than MODIS ² CMF difference 0. 02 OMI Science Team 2014, Anders Lindfors / FMI
OMI vs. MODIS (#7): CMF vs. latitude • Water clouds • 10 th/90 th percentile limits: COD 1— 80 • OMI CMF clearly higher than MODIS • Finnish latitudes (60 N): ² CMF difference 0. 06 OMI Science Team 2014, Anders Lindfors / FMI
To Conclude • Results are preliminary, more analysis needed: ² categorize by SZA, VZA, etc. ² regional aspects • OMI underestimates water cloud COD as compared to MODIS • OMI overestimates ice cloud COD as compared to MODIS • Overall: overestimation somewhat dominates ² can only explain 5— 10% of systematic difference between cloud-free and overcast surface UV ² At FMI’s stations observed difference is ~20 % • How good is MODIS? OMI Science Team 2014, Anders Lindfors / FMI
COD as function of wavelength • OMI COD is representative for UV wavelengths, based on radiance at ca 360 nm • MODIS is representative for midvisible, based on visible and IR radiances (what precisely? ) • Figure shows the COD of lib. Radtran following Hu & Stamnes – minimum tau=7. 44 (360 nm) – maximum tau=7. 65 (660 nm) • This means MODIS and OMI CODs are comparable although there is a different in wavelength
- Slides: 17