6- world meteorological organization（6 -世界气象组织）.docVIP

6. CALIBRATION AND VALIDATION Instrument calibration Introduction Calibration is the process of quantitatively defining the satellite instrument response to known, controlled signal inputs. The calibration information is contained in a calibration formula, or calibration coefficients that are then used to convert the instrument output (“counts”, previously “analogue signals”) into physical units (e.g., a radiance value). Instrument calibration is critical for any higher-level data processing, especially for deriving quantitative products or when data from different instruments need to be merged (e.g., composite imagery). For climate applications, the requirement for accurate calibration is particularly stringent since detection of small trends over long periods requires comparability of different instruments flown on different satellites at different times. Building homogeneous climate data records is contingent on very good calibration and error characterization. The following considerations apply to passive and active instruments alike. Five calibration domains should be generally considered: radiometric, spectral, spatial, temporal, and polarization. A complete calibration record should allow estimates of uncertainties in calibration parameters. Satellite instrument calibration should take into account all phases of an instrument’s lifetime: from design and pre-launch phases to post-launch, and on-orbit operations. Inter-calibration of instruments against a common reference instrument allows achieving consistency among satellite measurements at a given point in time. By comparing model-simulated and observed satellite radiances in data assimilation schemes, major NWP centres can also help determine relative biases between instruments. Calibration to absolute standards is, however, necessary to allow traceability of errors and to detect any long-term drift over time. Calibration using well-characterized, stable Earth targets (“vicarious” calibration