Feedback on Level-1 data from CCI projects R. Hollmann, Cloud_cci
Background Following this years CMUG meeting & Science Leader discussion on Level 1 CCI projects ingest a lot of level 1 satellite data and using a retrieval scheme to generate Level 2 ECV s (and level 3 as well) during validation and evaluation of level 2/ level 3 data it is possible to address changes in the product quality to imperfect level 1 data to collect and prepare a summary presentation what has been obtained from the CCI projects
Contribution of European EO Missions to climate records
Contribution of European EO Missions with other missions data
GOME instruments GOME Channel-tochannel jumps are caused by inhomogeneities in the observed scene during the serial read-out Inaccuracy problems by determining aerosol type in retrieval schema. Uncertainty in AOD for retrieved type of aerosol Update in correction coefficients for the spectral bands is required. GOME-2 Calibration issue in the GOME-2/ MetOp level-1 spectra: significant dependence with the satellite scan angle Inaccuracy problems by determining aerosol type in retrieval schema. Uncertainty in AOD for retrieved type of aerosol Determine a bias between the relative differences for the west and east GOME-2 pixels. Make corrections for the calibration coefficients Radiometric degradation of the instrument. Reduction of the AAI data quality. Possible artificial trends in the AAI. Problem is being worked by EUMETSAT. Possible remove trend by comparing to other instruments (OMI).
ASAR & SPOT VGT instruments ASAR SPOT VGT Level 1 of high quality except some outliers presenting missing lines, wrong calibration or discontinuities. Level 2 (daily surface reflectance) of high quality but poor cloud detection Over 200,000 ASAR images (IMM, WSM and GM1 modes) images processed for global water bodies mapping. Most images are well suited for use. Missing lines or repeated lines were encountered in early datasets. Wrong calibration was applied in some late data. Occasional drop of backscatter at edges of some images. For ScanSAR data, occasional intensity offsets between scans visible but effect negligible. Over 5110 global S1 images processed for time series analysis. Impressive co-registration over years (< 1 km) but improved cloud detection was required, developed and implemented.
OMI / GOMOS instruments OMI Relatively large pixel size results in cloud contamination. Biased in the AOD retrievals. Improve cloud screening. Include MODIS data for cloud screening. After 2008 about 1/3 rd of the swath is blocked. AAI and AOD retrievals limited to a reduced swath. Corrections for the blockage are under study by KNMI and NASA. GOMOS Residual scintillation Reduced accuracy of extinction profile Use of a spectral/full spatial inversion scheme, in combination with a statistical estimate of the residual scintillation component (the Full Covariance Matrix method). Considered in the frame of AERGOM.
SCIAMACHY instrument SCIAMACHY In the UV, an increasing degradation is visible which depends on wavelength, especially close to the channel edges and in channel 2. Biased in the accuracy of AOD and aerosol type retrievals. A degradation correction is required to assure the quality of the SCIAMACHY data products. Corrections are being worked.
AVHRR instruments AVHRR Degradation of visible channels and intercalibration The used FCDR has been based on visible calibration corrections provided by NOAA (Heidinger et al, 2010). No corrections were made for infrared channels. The validity of corrections was confirmed by comparison with MODIS radiances (Collection 5) for the period 2007-2009 (Karlsson and Johansson, 2014). An update of the visible part of the FCDR based on a revised methodology and with reference to MODIS Collection 6 (Heidinger, 2014, personal communication) became available in October 2014. Navigation problems for older satellites (prior to NOAA-15) Inconsistent model for infrared calibration Navigation accuracy for many NOAA satellites prior to the NOAA-KLMNN series of satellites is sometimes poor. Deviations of several GAC pixels (i.e., more than 20 km) have been observed. Clock corrections for afternoon satellites are available (solving most of the problems) but a solution for morning satellites is still needed.
MERIS instrument MERIS Level 1 of high quality except outliers regarding to the input data quality. formally incorrect files (shorter than specified in header, missing tie-point lat/lon) processing failed spectral campaigns shifted geocoding unexpected band data vertical stripes Over 131187 MERIS FR images and 48940 MERIS RR images are processed in the CCI Land Cover project. Around 99% of the images are very well suited for use. The use of the full mission dataset in a consistent way requires the development of techniques for the assessment of the input quality. Coverage limited to a lower latitude in winter time due to earlier cut off than other similar sensors (MODIS, SPOT VGT) in Northern hemisphere No data acquisition in late December from Brussels and northward, limiting the possible use for LC mapping and snow detection Negative bias for 0.8 micron channel reflectances Karlsson and Eriksson (2014) reported a negative bias of 3.5 % compared to corresponding MODIS channels.
AATSR instrument AATSR Positive bias for 0.6 micron channel reflectances A positive bias exceeding 5 % in reflectance observed using MODIS (Collection 5) radiances as reference (Karlsson and Eriksson, 2014). Could partly be caused by insufficient compensation for spectral response differences. Negative bias for 1.6 micron channel reflectances In the same study as referenced above a negative bias of 3.5 % was found. Problems in compensating for changes in cloud and surface spectral signatures might explain most of this.
MW-instruments Nimbus-7/ SMMR and Aqua/AMSR-E + GCOM-W1 / AMSR2 Bias between brightness temperatures The C-band radiometer SMMR entirely lacks an overlap in time with other C-band radiometers and shows only very little temporal overlap with radiometers operating at other frequencies (i.e. DMSP- SSM/I). This hampers a direct intercalibration between the sensors, which leads to biases in the brightness temperatures and hence in the retrieved soil moisture values. Request for a SMMR intercalibration Similar issues for: Aqua AMSR-E and GCOM-W1 AMSR2 ERS AMI-WS and MetOp ASCAT ERS-2 & ERS-1 AMI-WS
Summary / next steps Received feedback on level 1 from ~5 Cci projects good starting point Requirements for further inter-calibration Communication: projects reported their good experience with colleagues embedded in the project and the corresponding Quality Working group to receive the latest updates/ versions Official information path of ESA on updates and issues with Level-1 not timeliness survey needs to be continued