1. Assumption of Lambertian Cloud Surface (I)
TOMS Ozone Retrieval Sensitivity to
Assumption of Lambertian Cloud Surface (I)
Xiong Liu,1 Mike Newchurch,1 Robert Loughman2 , and Pawan K. Bhartia3
1. Department of Atmospheric Science, University of Alabama in Huntsville, Huntsville, Alabama, USA
2. Cooperative Center for Atmospheric Science & Technology, University of Arizona, Tucson, AZ, USA
3. NASA Goddard Space Flight Center, Greenbelt, Maryland, USA
Radiative Transfer Models
Treat clouds as scattering medium, calculate the backscattered radiance at the top of the atmosphere using Polarized Plane-parallel Gauss-Seidel Radiative Transfer Code (PPGSRAD) at N7 TOMS six channels (312, 317, 331, 339, 360, and 380 nm). Polarization is considered for clouds with optical depth  150.
Retrieve ozone using TOMS version-7 algorithm (TOMSV7), from which the look-up table is calculated using TOMRAD code at 10 pressure levels from 1.0 to 0.1 atm to reduce radiation interpolation error.
The radiance difference between TOMRAD and PPGSRAD is 0.2%, on average, for clear sky conditions.
Use wavelength-weighted ozone absorption coefficients, rayleigh scattering coefficients and molecular depolarization factor at each channel (consistent in PPGSRAD and TOMRAD).
TOMS standard low-latitude ozone profile L275 is used as the original profile.
Optical properties of water clouds are computed by Bohren-Huffman’s Mie Code. Optical properties of polycrystals and hexagon column crystals are computed by Ray Tracing Code.
Abstract. Using a radiative transfer model that treats clouds as scattering medium in the forward simulation, we study the assumption of opaque Lambertian cloudy surface and the employed PCM on TOMS ozone retrieval. The assumption of angularly independent cloud reflection is fairly good because the Ozone Retrieval Error (ORE) is within 1.5% of the total ozone when Cloud Optical Depth (COD)  20. Because of the In-Cloud Ozone Absorption ENhancement (ICOAEN), the assumption of opaque cloudy surface introduces large OREs even for optically thick clouds. For a water cloud of COD 40 at 2-12 km with 20.8 DU ozone homogeneously distributed in the cloud, the ORE is 17.8 DU at nadir view. This ICOAEN effect depends greatly on viewing geometry, ozone amount in the cloud, and ozone distribution in the cloud. The ICOAEN effect for those tropical high-reflectivity convective clouds (reflectivity  80% and cloud top pressure  300 hPa) is typically 5-13 DU over the Atlantic Ocean and Africa, and 1-7 DU over the Pacific Ocean. The TOMS PCM is good because negative errors from the cloud fraction being underestimated partly cancel positive errors due primarily to the ICOAEN effect. At COD  5, the PCM effect almost offsets the ICOAEN effect, and the TOMS algorithm retrieves the about correct TOC. With increasing COD up to 20-40, the negative PCM effect decreases more dramatically than the positive ICOAEN effect, so the overall positive ORE increases.
Methodology
Four effects on ozone retrieval: the assumption of angularly independent cloud reflection (Lambertian effect), the effect of using PCM (PCM effect), the effect of In-Cloud Ozone Absorption ENhancement (ICOAEN effect), Below Cloud-bottom Ozone Absorption (BCOA effect)
Lambertian effect: simulate radiance without ozone below clouds, retrieve ozone by forcing the cloud fraction to be the forward cloud fraction, the difference between retrieved ozone and forward input-ozone is the Lambertian effect.
PCM effect: same as above, but retrieve ozone using the TOMS PCM, then the difference in the retrieved ozone between using PCM and forcing the cloud fraction to be forward cloud fraction.
ICOAEN and BCOA effects: calculate radiance with and without ozone in the clouds (below cloud-bottoms), retrieve the total ozone for both radiances, the difference between two retrieved ozone is the ICOAEN ( BOCA) effect.
Viewing geometry: solar zenith angle (SZA) (0°, 15°, 30°, 45°, 60°, 70°, 75°) and view zenith angle (VZA) (0°-70° every 5°) , relative azimuthal angle (AZA) (0°-180° every 30°).
To represent those tropical high-reflecting clouds, a typical homogeneous water cloud is put between km with an COD of 40 (corresponding to cloud reflectivity of ~80%). Forward cloud fraction is assumed to be 1.
Investigate how these effects vary with cloud optical properties, cloud optical depth, cloud location, ozone amount and distribution in the clouds.
Motivation and Objectives
Cloud treatment in operational algorithms is highly idealized. TOMS V7 algorithm assumes clouds as opaque Lambertian surfaces and uses the partial cloud model (PCM) (Minimum Full cloud reflectivity is 80%). These idealizations may cause ozone retrieval errors.
The significant ozone excess of DU over tropical high-altitude, highly reflecting clouds compared to clear observations motivates to study the TOMS cloud treatment on ozone retrieval.
Possible sources of ozone retrieval errors are illustrated in Figure 1.
Cloud reflection is angularly dependent.
Photons penetrate into clouds even below cloud-bottom.
The derive cloud fraction using the PCM may be different from the actual.
We use radiative transfer codes to address the effects of these aspects on TOMS ozone retrieval for thick clouds.
Photons penetrate into the clouds, and photon path length is enhanced through multiple scattering, resulting in enhanced ozone absorption.
Non-Lambertian cloud surface
Incident light
Tropopause
Ice clouds
A TOMS partial cloud scene might be total cloudy with R < 80% (e.g., only 40%) or might be broken clouds.
Water clouds
Cloudy
Partially Cloudy
Ground Surface
Figure 1. Possible sources of ozone retrieval errors. 1 2
Table 1. The range and average of the Lambertian-PCM effect.
Lambertian-PCM effect
The Lambertian effect is mainly due to the difference in the ozone absorption enhancement resulting from Rayleigh scattering and cloud reflection between simulated scattering clouds in PPGSRAD and assumed Lambertian clouds in TOMRAD. That is why it varies with cloud-top height, cloud optical thickness, ozone profiles, and phase function.
At COD  20, the Lambertian-PCM effect is typically within 1.5% of TOMS TOC (Table 1), approximately within the accuracy of TOMS ozone retrieval, indicating assuming cloud scattering as isotropic is fairly good.
Figure 2. Sum of Lambertian and PCM effects for water clouds at OD=40 (left) and OD=10 (right). * indicates the SZA.
Figure 3. (a) Lambertian-PCM effect, (b) Lambertian effect, and (c) PCM effect for COD 40 (left) and COD 10 (right).
Lambertian and PCM Effects ( or Lambertian-PCM effect)
Figure 2 shows the sum of Lambertian and PCM effects as a function of viewing geometry for a water cloud of COD 40 (left) and 10 (right) at 2-12 km. The cloud fraction in the forward cloud fraction
The Lambertian-PCM effect varies with viewing geometry. At COD 40, the error is within ±4.5 DU. The error at COD 10 ranges from –8.9 DU to 4.4 DU.
Figure 3 separates the Lambertian effect from the PCM effect. The error caused by the Lambertian effect is slightly more scattered for COD 40 than for COD 10. The negative PCM effect is due to mainly to the cloud fraction being underestimated so the added ozone below clouds is reduced.
With decreasing COD, the negative PCM effect increases in magnitude because the effective cloud fraction decreases, and the Lambertian-PCM effect becomes dominated by the negative PCM effect.
Table 1 shows the range and average error due to the Lambertian-PCM effect.
The Lambertian-PCM effect varies with cloud optical depth, cloud optical properties (Water Clouds with Henyey-Greenstein phase function (WCHG), HEXagon ice crystals (HEX), POLYcrystals (POLY) ), ozone profiles, and cloud-top pressure.
ICOAEN Effect vs. Viewing Geometry
Figure 4 shows the ICOAEN effect vs. viewing geometry for a water cloud of COD 40 positioned at 2-12 km. There is 20.8 DU ozone in the cloud.
The ICOAEN effect decreases dramatically with the increase of SZA and VZA and is azimuthally independent, ~18 DU at nadir and only 0.15 DU at SZA = 75° and VZA = 70°. The exchange of SZA and VZA does not change ICOAEN effect.
The photon path length in clouds decreases with increasing SZA and VZA. Furthermore, TOMS algorithm automatically accounts for the geometrical path length (1/cos(SZA) + 1/cos(VZA)). These two factors lead to the dramatic decrease of enhanced ozone vs. geometrical path length.
Figure 4. ICOAEN effect vs. Viewing Geometry. * indicates the SZA. 3 4
Updated on August 30, 2002