Skip to main content
CERES Home Page CERES Home Page

CERES-MODIS cloud retrieval algorithm

The CERES-MODIS cloud retrieval algorithm is based on (Minnis et al. 2009) and are not the Goddard official product. The CERES-cloud Edition3 mask designates all MODIS pixels as either clear or cloudy. Cloud property retrievals are performed on all cloudy pixels. The cloud properties are used to convert CERES radiances into fluxes using scene specific CERES ADMs. Un-retrieved cloudy pixels will cause biases in the ADM selection. The Terra and Aqua MODIS cloud properties for Edition3 are consistent. Both Aqua and Terra use the same set of MODIS channels to derive cloud properties, for example the 2.1µm channel is used in place of the 1.6µm. The daytime cloud retrievals are based on the visible, 3.7µm and IR MODIS channels and are more reliable than the nighttime cloud retrievals only based on the IR MODIS channels.

Multi-layer clouds

Multi-layer clouds are identified if the CO2 slicing cloud height is much greater than the standard retrieval. It is assumed that the there is a high thin cloud over a low thick cloud. Cloud retrievals are recomputed based on the CO2 cloud height and the original optical properties are distributed into two layers and are given as new Edition3 SSF cloud parameters. Only the cloud fractions for the overlap categories are saved for the hourly, daily and monthly datasets. The overlap properties will be spatially and temporally averaged into the CERES ISCCP-D2like products.

CERES cloud overlap fraction index

Overlap index
(x) indicates cloud
1 2 3 4 5 6 7 8 9 10
300mb - 70mb       X X X X      
500mb - 300mb       X   X     X  
700mb - 500mb     X       X     X
Surface - 700mb   X           X X X

CERES cloud layers

There are 5 CERES regional static cloud layers stratified by effective pressure level as indicated in the table below.

Cloud Layer Pressure level
Total Surface - 70mb
High 300mb - 70mb
Upper middle 500mb - 300mb
Lower middle 700mb - 500mb
Low Surface - 700mb

CERES-MODIS cloud property spatial averaging

All MODIS pixels are identified as either clear or cloudy inside a given CERES 1.0° gridded region. The regional cloud property is essentially the average of the MODIS pixel level cloud properties. The cloud fraction is the fraction of cloud pixels divided by total pixels.

Footprint averaged cloud properties

The MODIS pixels are convolved into a CERES footprint (20km nadir). Each CERES footprint provides a single flux (SSF) measurement, the cloud properties are used to convert the SSF radiance into a flux. The cloudy pixels are stratified into up to 2 dynamic cloud layers, where each cloud layer property is the average of the pixel level properties for the layer. The pixels are weighted by the footprint optics throughput known as the point spread function (PSF). The total footprint cloud amount is the summation of all the footprint layered cloud amounts. The footprint cloud layers are then assigned to the static CERES cloud layers. The mean total footprint cloud property is the layer fraction weighted cloud property. The mean total optical depth is averaged in terms of log optical depth.

CERES regionally gridded cloud properties

The CERES uses a 1.0° latitude by 1.0° longitude spatial grid. The CERES footprints whose center latitude longitude that fall within the given region are then equally weighted to derive the regional cloud properties. Only the footprints with valid CERES fluxes are used. The layered regional cloud amount is the average of the footprint layered cloud amounts. The layered regional cloud property is the footprint layered cloud amount weighted cloud property. The cloud optical depth is averaged in terms of log optical depth.

CERES Data Product Information

  Image: NASA Logo NASA Official: Dr. Norman Loeb
Page Curator: Edward Kizer
Page Last Modified: 10/16/2017 11:04:36 EST
Site Last Modified: 11/07/2017 17:18:43 EST