Evaluation Dataset for Passive Imager Retrievals
The Cloud Retrieval Evaluation Workshops (CREWs) provide a forum for international satellite-based cloud retrieval teams to share their experience with state-of-art cloud parameter retrievals from passive imaging satellite observations. The evaluation of passive imager cloud properties retrievals is part of the CREW. A Common Database was built to organize cloud property retrievals from different algorithms from research institutes from Europe and America for passive imagers (MSG, MODIS, AVHRR, POLDER and/or AIRS), complemented with cloud measurements that serve as a reference (CLOUDSAT, CALIPSO, AMSU, MISR) for a number of “golden days”. All CREW participants are cordially welcome to use the datasets for their own studies. Due to our data policy, it is necessary to register in order to have access to the database. It is asked that you offer the institute a co-authorship, if you use their datasets for publications.
Note: the selected Golden Days are similar to those selected for CREW-2. Thus, providers that already have provided their data for CREW-2, and that do not want to update their datasets or add new ones, do not need to upload their datasets again.
The Common Database, which contains cloud parameter data sets from research institutes from Europe and the USA, was created for the evaluation of cloud properties retrievals in the CREW project. Table 3 lists the provider, the cloud parameters, and the satellites of the 16 data sets in the CREW Common Database. A detailed description of the retrieval algorithms can be found in the
Algorithm discription (only for CREW members, password needed).
Table1: List of Research groups that provided Cloud Parameter retrievals for the Common Database. For each groups the satellite instruments and cloud parameters for which data are provided are given. Where the codes from the cloud parameters are: cmk= cloud mask; cph= cloud phase;ctt= cloud top temperature; ctp= cloud top pressure; cth= cloud top height; cod = cloud optical depth; ref= effective radius; lwp= liquid water path; iwp= ice water path; cty= cloud type, and cs= convective signature. The coverage F means Full disc and S means swath data along the satellite path. COCS makes retrievals for ice clouds only (cloud phase = I) and the retrieved optical thickness is limited those smaller than 3 (cloud phase = F*). The EIM algorithm makes retrievals for water clouds only (cloud phase = W).
|CMS||DWD||Stenger, Martin; Lockhoff, Maarit; Kniffka, Anke||SEVIRI||F||F||F||F||F||F||F||F||F|
|EIM||Uni Marburg||Kühnlein, Meike||SEVIRI||F||W||F||F|
|FUB||FU Berlin||Preusker, Rene||SEVIRI||F||F|
|GSF||Nasa Goddard||Platnick, Steve||SEVIRI||F||F||F||F||F||F||F||F|
|GSF||Nasa Goddard||Platnick, Steve||MODIS||S||S||S||S||S||S||S||S||S|
|JMA||Japan Met. Agency||Takeuchi, Ayako||SEVIRI||S||S||S||S||S|
|LAR||Nasa Langley||Minnis, Patrick||SEVIRI||F||F||F||F||F||F||F||F||F|
|LAR||Nasa Langley||Minnis, Patrick||MODIS||S||S||S||S||S||S||S||S||S|
|MFR||Meteo France||Le Gleau, Herve; Derrien, Marcel||SEVIRI||F||F||F||F||F|
|UKM||UK Met Office||Francis, Pete||SEVIRI||F||F||F||F||F||F|
|ULI||Univ. Lille||Riedi, Jerome||POLDER||S||S|
All data are available on the CREW FTP site (registration needed).
The updated SEVIRI retrievals (in the same format for all groups) of CREW-3 are available in L2_CREW3_SEVIRI_COM.
All submitted data (all sensors) in the individual format of the group are available in L2_CREW3_ORG.
Table2: List of Research groups that provided Reference data for the Common Database. For each groups the satellite instruments and cloud parameters for which reference data are provided are given. The coverage P means profiles and S means swath data along the satellite path.
*) Only available for single layer ice clouds
**) Only available for single layer water clouds
We have selected five Golden Days that are listed in Table 1. During several hours of the selected Golden Days the A-Train satellites were aligned with AVHRR on-board NOAA-18. Our evaluation study will focus on these hours. However, we encourage the data providers to upload their data sets of the five Golden Days for the full 24 hrs. The study area is restricted to the SEVIRI full disk. The following data are desired for the Common Database:
- SEVIRI 15 minute data of the MSG disk preferably covering the full 24 hrs during the golden days (5x24hrs);
- MODIS swath data covering the MSG disk preferably covering the full 24 hrs during the golden days (5x24hrs);
- AVHRR swath data covering the MSG disk during the hours where NOAA-18 was aligned with the A-Train satellites
- Other instruments (MERIS/POLDER/..) preferably covering the full 24 hrs during the golden days (5x24hrs).
You can upload your data to the Common Database through the FTP server at ICARE, which is assessable for CREW members after registration. During the registration when asked for a “short description of your project”, please write: Account request for the Cloud Retrieval Evaluation Workshop (CREW).
Table 4: Golden Days that have been selected for the (inter)comparison
|Day||Month||Year||Day of Year||Hours with alignment between A-Train and NOAA-18||A-Train orbit numbers|
|13||June||2008||165||12:00-15:30||11317, 11318, 11319|
|03||July||2008||185||10:00-12:00||11607, 11608, 11609|
The inter-comparison of cloud parameters is done for level 2 products. The products delivered by the different algorithm providers have been converted to a common file format to facilitate direct inter-comparison. Table 2 lists the cloud parameters have been selected for the (inter)comparison:
Table 5: Cloud Parameters that have been selected for the (inter)comparison
|ctt||cloud top temperature|
|ctp||cloud top pressure|
|cth||cloud top height|
|cod||cloud optical depth|
|lwp||liquid water path|
|iwp||ice water path|
The CREW evaluation activities concentrate on passive imager instruments onboard both polar and geostationary satellites (SEVIRI, AVHRR, and MODIS). Furthermore some participants also use polarized radiances of POLDER, the spectral information of MERES, and the multi angle observation of MISR. The active sensors CALIOP and CPR and the micro waver intrument AMSR-E are used as reference for the retrievals using passive sensors.
SEVIRI is a 50 cm-diameter aperture, line-by-line scanning radiometer, which provides image data in four Visible and Near-InfraRed (VNIR) channels and eight InfraRed (IR) channels. A key feature of this imaging instrument (Fig. 1) is its continuous imaging of the Earth in 12 spectral channels with a baseline repeat cycle of 15 min. The imaging sampling distance is 3 km at the sub-satellite point for standard channels, and down to 1 km for the High Resolution Visible (HRV) channel.
MODIS (Moderate-resolution Imaging Spectroradiometer) is a payload scientific instrument launched into Earth orbit by NASA in 1999 on board the Terra (EOS AM) Satellite, and in 2002 on board the Aqua (EOS PM) satellite. The instruments capture data in 36 spectral bands ranging in wavelength from 0.4 µm to 14.4 µm and at varying spatial resolutions (2 bands at 250 m, 5 bands at 500 m and 29 bands at 1 km). Together the instruments image the entire Earth every 1 to 2 days.
A description of the MODIS instrument and access to the dataset is given at the official MODIS website.
The AVHRR instrument (Advanced Very High Resolution Radiometer) measure the reflectance of the Earth in 5 relatively wide (by today's standards) spectral bands. The first two are centered around the red (0.6 micrometer, 0.5 THz) and near-infrared (0.9 micrometer, 0.3 THz) regions, the third one is located around 3.5 micrometer, and the last two sample the thermal radiation emitted by the planet, around 11 and 12 micrometers, respectively.
POLDER (POLarization and Directionality of the Earth's Reflectances) is a wide field of view imaging radiometer that has provided the first global, systematic measurements of spectral, directional and polarized characteristics of the solar radiation reflected by the Earth/atmosphere system. Its original observation capabilities have opened up new perspectives for discriminating the radiation scattered in the atmosphere from the radiation actually reflected by the surface.
MERIS (The Medium Resolution Imaging Spectrometer Instrument) is a wide field-of-view pushbroom imaging spectrometer with a swath width of 1150km (field-of-view (FOV) = 68.5°) measuring the solar radiation reflected by the Earth in 15 spectral bands in the solar range from about 412.5nm to 900nm. All bands are programmable in width (variable between 1.25 and 30 nm) and position, but are fixed before launch in response to the recommendations of the Science Advisory Group (SAG) for the main period of the mission. The instrument scans the Earth's surface by the so called "push-broom" method. Linear CCD arrays provide spatial sampling in the across-track direction, while the satellite's motion provides scanning in the along-track direction. Each MERIS pixel has a field of view of 0.019°. Due to the wide instrument field of view (68.5°), spatial sampling varies in the across track direction, between 0.26 km at nadir and 0.39 km at swath extremities. Along-track sampling is close to 0.29 km.
Viewing the sunlit Earth simultaneously at nine widely spaced angles, MISR (Multi-angle Imaging SpectroRadiometer) provides ongoing global coverage with high spatial detail. In each of the nine MISR cameras, images will be obtained in four spectral bands, i.e. in four different colors, one each for blue, green, red, and near-infrared. The center wavelength of each of these bands is 446, 558, 672, and 867 nanometers respectively. Each MISR camera sees instantaneously a single row of pixels at right angles to the ground track. MISR collects data only on the daylit side of the Earth. During each orbit, MISR obtains a swath of imagery that is 360 km wide by about 20,000 km long. The intrinsic crosstrack dimensions of the MISR pixels was therefore chosen to be 275 meters at all off-nadir angles. The nadir camera makes use of the A camera design, resulting in a slightly higher crosstrack resolution of 250 meters, and this allows it to provide a slightly better ground locating reference that is passed on to the observations from the other cameras.
A-TRAIN reference instruments
CALIOP (Cloud-Aerosol LIdar with Orthogonal Polarization) [Anderson, et al., 2005] is the primary instrument on the CALIPSO satellite. It is a dual wavelength polarized Lidar launched in April 2006 on board CALIPSO. Measurements are taken at two wavelengths, 532 nm and 1064 nm. Primary products are the profiles of total backscatter, as well as profiles of cloud and aerosol properties. The instrument also measures the linear depolarization of the backscattered return at 532 nm, allowing discrimination of cloud phase and the identification of the presence of non-spherical aerosols. Individual CALIOP beams have a width of about 70 meters on the Earth’s surface with a sampling distance of 333 m on the surface. Products are available at different resolutions with a footprint size of typically 1-5 km and a vertical resolution of 30-60 meters (nominal). Its high sensitivity and vertical resolution make CALIOP an excellent candidate system for the analysis of cloud top height as well as cloud phase (near top). Additionally, its high sensitivity to aerosols makes it a valuable tool to detect aerosol layers above clouds. As regards water clouds, CALIOP saturates fairly quickly near the cloud top and will not provide much information about the inner structure of the clouds. However, for optically less dense ice clouds, CALIOP provides information about ice also from deeper within the cloud [Chiriaco, et al., 2007].
CALIOP observation and retrieval products are available on the ICARE FTP site: CALIOP.
CPR (Cloud Profiling Radar) onboard of Cloudsat is a 94-GHz nadir-looking radar was lanched in April 2006. It measures the power backscattered by clouds, scene classifications, and cloud optical properties as a function of distance from the radar with a vertical resolution of 500 m, a cross-track resolution of 1.4 km, and an along-track resolution of 1.7 km.
Cloudsat CPR observation and retrieval products are available on the ICARE FTP site: CLOUDSAT.
AMSR-E (Advanced Microwave Scanning Radiometer for EOS) is a twelve-channel, six-frequency, total power passive-microwave radiometer system. It measures brightness temperatures at 6.925, 10.65, 18.7, 23.8, 36.5, and 89.0 GHz. Vertically and horizontally polarized measurements are taken at all channels. The Earth-emitted microwave radiation is collected by an offset parabolic reflector 1.6 meters in diameter that scans across the Earth along an imaginary conical surface, maintaining a constant Earth incidence angle of 55° and providing a swath width array of six feedhorns which then carry the radiation to radiometers for measurement. Calibration is accomplished with observations of cosmic background radiation and an on-board warm target. Spatial resolution of the individual measurements varies from 5.4 km at 89.0 GHz to 56 km at 6.9 GHz.
Brightness temperatures of AMSR are available of the ICARE FTP site: AMSR_E.
Instead of using real observations, it is possible to simulate the observations with a radiative transfer model. The advantage of these simulated satellites scenes is, that the state of the atmosphere and the clouds is known. In this way the results of the retrieval can be diretly compared to the 'known' atmospheric state. The disadvantage is, that assumptions and uncertainties may occur in the forward simulation, including the horizontal homogeneity, vertical structued clouds, three dimensional radiative transfer, and not perfectly known optical properties of the surface, gaseous absorption, and clouds. Noting deviations of the retrieval result and the 'known' atmospheric state, it is always nessesary to check weather the deviation is caused by an imprecise retrieval algorithm or an inaccuracy of the forward model. Using the same forward model and assumptions for the simulations of satellite observations and for training the retrieval algorithm may also lead to a situation, where inaccuracy of radiative transfer and retrieval cancel each other out. Nevertheless this approach is useful as plausibility check, to identify to rough assumptions in the retrieval algorithm, and to distinguish properties that may be retrieved and that are impossible to retrieve. Simulations may be performed with different radiative transfer models for different sensors. The most known radiative transfer models are libRadtran, S6, and RTTOV and more.
Upload Instructions for Dataset Providers
Uploading instructions for cloud remote sensing datasets providers.
For questions related to access and use of ICARE ressources send comments or questions to
Rob Roebeling, Jerome Riedi and Ulrich Hamann