Comparison of different cloud fraction algorithms for OMI

D. Loyola (DLR), O. Torres (NASA), J. Joiner (NASA), P. Veefkind (KNMI), PK. Bhartia (NASA),

D. Haffner (NASA), and R. Lutz (DLR)

OMI Science Team Meeting, September 12th 2017

Chart 2

Overview

Cloud fraction types and algorithms OCRA AERUV CLDO2 CLDRR TO3CRF MODIS

Global comparisons

Latitudinal comparisons

Summary

Chart 3

OCRA

The UV/VIS/(NIR) measurements are projected to the (R)GB color space

A radiometric cloud fraction is computed as the distance between RGB measurements and a cloud-free composite

No assumptions on cloud model No special handling of snow/ice

conditions

References: V1 GOME and SCIAMACHY: Loyola et al., TGRS 2007 V2 GOME-2: Lutz et al., AMT 2016 V3 OMI and TROPOMI: Loyola et al., AMTD 2017

Chart 4

AERUV

A radiatively effective cloud fraction at 388nm is calculated as:

𝐶𝐶𝑓𝑓 = (𝐼𝐼𝑜𝑜𝑜𝑜𝑜𝑜 − 𝐼𝐼𝑜𝑜𝑠𝑠𝑠𝑠)/(𝐼𝐼𝑐𝑐𝑐𝑐𝑜𝑜𝑠𝑠𝑐𝑐 − 𝐼𝐼𝑜𝑜𝑠𝑠𝑠𝑠) using a Mie water cloud model

with prescribed cloud optical depth of 10.

Surface albedo from OMI long-term minimum LER corrected with an ocean model

Cloud fraction is 0 for snow-ice for terrain press level < 600 hP

References for OMAERUV 1.8.9.1 Torres, O., P.K. Bhartia, H. Jethva, and C. Ahn, 2017: Impact of the

Ozone Monitoring Instrument Row Anomaly on the Long-term Record of Aerosol Products, AMT Discussions, to be submitted.

Chart 5

CLDO2

OMCLDO2 is based on fitting of the spectral range between 460 and 490 nm.

The cloud model is an opaque Lambertian surface with an albedo of 0.8.

Surface albedo is taken from the Kleipool climatology, and adjusted using snow-ice information.

The v2 OMCLDO2 product contains effective cloud fraction, effective cloud pressure, scene albedo and scene pressure.

Veefkind et al.: AMT, 9, 6035-6049, https://doi.org/10.5194/amt-9-6035-2016, 2016.

Chart 6

CLDRR

UV channel at 354.1 nm is used (no Raman) for cloud fraction.

Uses Cox-Munk with TOMS-based water leaving radiance over ocean for surface reflectivity, TOMS-based over land.

An effective cloud fraction is computed using the Mixed Lambertian concept assuming clouds are opaque with reflectivity > 0.8.

Over snow/ice conditions, cloud fraction set to 1, scene pressure retrieved. OMCLDRR references:

Vasilkov et al., JGR 2008 Vasilkov et al., AMT 2010 Joiner et al., AMT 2012

Chart 7

TO3CRF

Cloud radiance fraction (CRF) is estimated fraction of the measured radiance signal reflected by clouds in the instrument field-of-view.

The TOMS CRF is a simplified linear approximation of CRF.

CRF set to 0 for snow/ice conditions Haffner and Bhartia (2017) Total

Ozone Mapping Spectrometer Version 9 Algorithm, in preparation.

Chart 8

MODIS

MODIS geometric cloud fraction (derived from a variety of checks using MODIS wavelengths from visible to IR) averaged over OMI pixels

No assumptions about cloud model

Over snow, NIR channels are used to detect clouds with other channels

Aqua MODIS collection 5 used in this work (MYD05)

Chart 9

MODIS – OCRA

Chart 10

MODIS – AERUV

Chart 11

MODIS – CLDO2

Chart 12

MODIS – CLDRR

Chart 13

MODIS – TO3CF

Chart 14

Correlation and Histogram 60S-60N for 2005-09

Chart 15

Correlation and Histogram 60S-60N for 2005-09

Chart 16

Correlation and Histogram 60S-60N for 2005-09

Chart 17

Seasonal Latitudinal Mean 2005

Chart 18

Seasonal Latitudinal Mean – Land

Chart 19

Seasonal Latitudinal Mean – Sea

Chart 20

Summary

Five cloud fraction (CF) algorithms from OMI compared with MODIS Dissimilar CF type, wavelength, algorithms and climatology

Similar spatial patterns

Except for snow/ice conditions and elevated terrain The CF differences change over land/sea

Statistics for 60S- 60N in 2005

Data Set Mean ± StDev r Slope/Offset MODIS 0.644 ±0.183

TO3CRF 0.522 ± 0.168 0.92 0.84x + 0.00 AERUV 0.423 ± 0.171 0.87 0.78x - 0.07 OCRA 0.360 ± 0.154 0.89 0.75x - 0.10

CLDRR 0.294 ± 0.148 0.80 0.56x - 0.08 CLDO2 0.280 ± 0.127 0.85 0.58x - 0.08