The ATSR Project
Processing of the ATSR-2 visible calibration tables has been transferred from a Vax station 4000 to an Alpha running VMS. To reflect this change the file versions have been revised to V1.4 for the uncorrected table and V2.1 for the corrected table. The calibration algorithms are unaffected. All users should check and update any relevant bookmarks so that they get the correct files.
Users must note that as from July 1999 the calibration table has been corrected for long-term drift and a bias in the 1.6µm calibration level. For those who have already included these corrections or do not wish to reprocess their data, an uncorrected version of the table will still be maintained at this site.
How to calibrate ATSR-2 Reflectance Images
Data from the reflectance channels of ATSR-2 are un-calibrated in SADIST-2 products created by the current version of the software. The data in such products have the unit "normalized count" as described in ref. .
To convert these data to top-of-atmosphere fractional-reflectance units requires only linear scaling by a single conversion factor (one-per-channel). These conversion factors have been derived using data from the on-board visible calibration system (VISCAL) which is described in detail in ref. . Users can convert their data from counts to reflectance using a calibration Look-Up-Table. The LUT consists of a time series of conversion factors; each record in the table consists of (from left-to-right):
The "SADIST time" is the "acquisition time" in the product name - as explained on page 58 of ref. , and is in the form YYMMDDhhmm. E.g. 9506010652 means the calibration refers to a measurement made at 0652h on June 1st, 1995. The simplest way to use the table is to locate the SADIST acquisition time closest to that in the name of any product in question and then select the conversion factor, F, for the appropriate channel. The conversion from counts, C, to top-of-atmosphere reflectance, r, is then
where, µ0 is the cosine of the solar zenith angle, and can be derived from the information in the SADIST-2 product header. L is the detected radiance and I0 is the solar irradiance. Because the calibration system uses the Sun as its source, no correction factor is required for sun earth distance.
Note: If the conversion factor for any channel is 0.00000000, this indicates that no valid measurement is available for this point in time. The next nearest time record must be used instead.
As with any optical system, there will be degradation of the optical surfaces due to high energy particles and UV radiation, resulting in a gradual drift of the calibration level. Using data from a number of large, spatially uniform and temporally stable sites, we have been able to determine the calibration drifts (see ref. ). The analysis yielded a drift function of the form
where t is the number of days since launch on 28th Apr 1995. The drift rates per year, k, for each channel are given in the table below.
The corrected calibration factor for each channel is
Various in-situ campaigns and vicarious analysis methods have been employed to verify the accuracy of the ATSR-2 visible channels. Comparisons between the ATSR-2 measurements and the expected reflectances show that the calibrations of the 0.87µm, 0.66µm and 0.56µm channels are within the stated accuracy (±3%). However, there is a consistent bias at 1.6µm that showed significantly higher than expected reflectances. The average bias factor from all of the validation methods was 1.06. The corrected 1.6µm calibration (including drift) is therefore
 Smith D.L., Read P.D. and Mutlow C.T., 1997, 'The Calibration of the Visible/Near Infra-Red Channels of the Along-Track Scanning Radiometer-2 (ATSR-2)' in Sensors, Systems and Next-Generation Satellites, Hiroyuki Fujisadsa, Editor, Proceedings of SPIE Vol. 3221, 53-62
Page last modified : Tuesday, 04-Dec-07