OMI Science Team 2014, Anders Lindfors / FMI 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
Feb 24, 2016
OMI Science Team 2014, Anders Lindfors / FMI
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
What is known about OMUVB performance?
• OMUVB is known to overestimate the surface UV
• Discussion has concentrated on aerosols as the reason for overestimation
• Mikko Pitkänen (MSc, 2013) comparison in Jokioinen and
Sodankylä, matching the overpass time
cloud classification using sunshine duration, cloud amount, surface solar radiation
OMUVB performance depends on clouds
overcast conditions: stronger overestimation
similar results also in other studies: Weihs et al. (ACP, 2008)
OMI Science Team 2014, Anders Lindfors / FMI
Sodankyläcloud-free
rMB = 0.08
Sodankyläovercast
rMB = 0.29
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
OMI Science Team 2014, Anders Lindfors / FMI
http://en.wikipedia.org/wiki/A-train_(satellite_constellation)
Matching OMI and MODIS CODsOMI 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
COD1 COD2
CMF1 CMF2• exponential relation R vs COD• logarithmic average of COD has been
found to be useful
• from MODIS cmp/w OMI COD
R2
R1
Figure from Zinner and Mayer (JGR, 2006)
MODIS
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
COD1 COD2
• exponential relation R vs COD• logarithmic average of COD has been
found to be useful
• from MODIS cmp/w OMI COD
R1,2
Figure from Zinner and Mayer (JGR, 2006)
OMI
CMF1,2
CMF = Cloud Modification Factor• CMF = Fall-sky / Fcloudfree
• CMF can be averaged (assuming independent pixel radiative transfer):
CMF1,2 = (CMF1 + CMF2)/2
CMFMODIS = CMF1,2,…,N
CMFMODIS cmp/w CMFOMI
• radiative transfer model used to calculate CMFMODIS and CMFOMI
OMI Science Team 2014, Anders Lindfors / FMI
COD1 COD2
CMF1,2 = ( CMF1 + CMF2 ) / 2
OMI vs. MODIS (#1): nr of colocated pixels
OMI Science Team 2014, Anders Lindfors / FMI
• 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 vs. MODIS (#2): COD vs. exponent of log-averaged COD
OMI Science Team 2014, Anders Lindfors / FMI
• All cases included
• 1.4 x 106
colocations• good agreement• OMI somewhat
lower than MODIS for COD>10
OMI vs. MODIS (#3): COD vs. exponent of log-averaged COD
OMI Science Team 2014, Anders Lindfors / FMI
• MODIS ice clouds• 500 x 103
colocations• OMI COD
somewhat higher than MODIS
OMI vs. MODIS (#4): COD vs. exponent of log-averaged COD
OMI Science Team 2014, Anders Lindfors / FMI
• MODIS water clouds
• 450 x 103
colocations• OMI COD clearly
lower than MODIS
Undestanding difference between ice and water clouds
OMI Science Team 2014, Anders Lindfors / FMI
ICE WATER
OMIOMI
More backscatter for same optical depth
• OMI cloud model always assumes water clouds
• Scattering phase function of ice: more backscatter
OMI sees ice clouds as thicker!
This explains relative difference between water / ice cloud performance
OMI vs. MODIS (#5): CMF vs. latitude
OMI Science Team 2014, Anders Lindfors / FMI
• All cloud types• 10th/90th percentile
limits: COD 1—80 • OMI CMF higher or
at same level as MODIS
• Finnish latitudes (60 N): small CMF
difference of 0.02—0.03
OMI vs. MODIS (#6): CMF vs. latitude
OMI Science Team 2014, Anders Lindfors / FMI
• Ice clouds• 10th/90th percentile
limits: COD 1—80 • OMI CMF lower than
MODIS CMF difference 0.02
OMI vs. MODIS (#7): CMF vs. latitude
OMI Science Team 2014, Anders Lindfors / FMI
• Water clouds• 10th/90th percentile
limits: COD 1—80 • OMI CMF clearly
higher than MODIS• Finnish latitudes
(60N): CMF difference 0.06
OMI Science Team 2014, Anders Lindfors / FMI
To Conclude• Results are preliminary, more analysis needed:
categorize by SZA, VZA, etc. regional aspects
• OMI underestimates water cloud COD as compared to MODIS
• OMI overestimates ice cloud COD as compared to MODIS
• Overall: overestimation somewhat dominates can only explain 5—10% of systematic difference between
cloud-free and overcast surface UV At FMI’s stations observed difference is ~20 %
• How good is MODIS?
COD as function of wavelength• OMI COD is representative for UV
wavelengths, based on radiance at ca 360 nm
• MODIS is representative for mid-visible, based on visible and IR radiances (what precisely?)
• Figure shows the COD of libRadtran following Hu & Stamnes– minimum tau=7.44 (360nm)– maximum tau=7.65 (660nm)
• This means MODIS and OMI CODs are comparable although there is a different in wavelength