Quality control of level1 satellite data
To be fully useful for weather, climate and environmental applications, satellite observations must be qualitatively and quantitatively controlled during the instruments lifetime: any radiometric systematic error not identified in the level1 radiances may propagate as errors in the retrieved variables. At LMD, the technique for inter-calibration has been initially developed for the calibration of Meteosat, based on space and time collocations with instruments on the NOAA series (J. Appl. Meteor., vol 21, 1982)
Based on work with TOVS (NOAA/NASA Pathfinder Programme), ATOVS, AIRS, IASI and IIR two complementary approaches: (i) an intercalibration approach and (ii) a "stand alone" approach have been developed which aim at identifying, and eventually at correcting, deviations or trends (natural, spurious) between pairs of channels of different instruments (in LEO/LEO or GEO/LEO modes).
To identify deviations or trends between pairs of channels, companion channels are selected based on the coherence of their radiative transfer properties as well as on the characteristics of their space, time, spatial resolution, viewing geometry. Sensitivity studies of all spectral channels to various observation conditions are performed using the radiative transfer tools derived, maintained and regularly validated at LMD: 4A/OP, GEISA, TIGR, ARSA, the high spectral resolution emissivity database.
Concerning IASI, such similar channels exist in both AIRS/Aqua, HIRS4/Metop, Tanso/Gosat. Concerning IIR/Calipso, such channels exist in Modis/Aqua or Seviri/MSG.
The inter-calibration approach is based on channel-by-channel comparisons between observations and observations made by other instruments (see below).
The stand alone approach is based on comparisons between observed and simulated radiances. Simulated radiances result from a forward radiative transfer model fed either with in situ radiosonde data or with products of reanalysis runs in collocation (time and space) with clear sky satellite observations. In our case the 4A/OP model and the ARSA database are used (see below).
The two approaches are complementary: the inter-calibration approach - not restricted to clear scenes - allows wide ranges of brightness temperatures being compared. and studies the behaviour of one channel relative to its companion. The 'stand alone' screens each channel of each instrument, individually, allowing e.g. the detection of viewing angle dependence of the brightness temperatures. When combined with the use of two companion instruments this contributes identifying which instrument deviates from the other(s).
Results for biases, standard deviations, trends, anomalies are reported in tables and/or plots. They are presented for 20K wide bins of temperatures ranging from 200 to 320K and for several observations conditions - e.g., latitude, day, night, viewing angles, etc...
Recent applications are within the frame of the validation of level1b of IIR/Calipso (Scott et al., Global Space-Based Intercalibration System, Gsics Quaterly, 2009, Vol3, No3) and IASI/AMSU/MetopA . Both actions are supported by CNES.