P5.29 The Calibration of NOAA AVHRR Visible Radiances with VIRS David R. Doelling, Venkatesan Chakrapani Analytical Services & Materials Inc., Hampton, VA Patrick Minnis and Louis Nguyen NASA Langley Research Center, Hampton, VA 1. INTRODUCTION There is a long history of Advanced Very High Resolution Radiometer (AVHRR) derived products that rely on accurate visible channel radiances. Currently AVHRR instruments lack onboard calibration for channels 1 (0.65 μm) and 2 (0.87 μm). Operationally, these channels are calibrated using a stable desert surface target as a reference to account for degradation of the sensors (Rao and Chen 1999 and 1996; http://psbsgi1.nesdis.noaa.gov:8080/EBB/ml/niccal.ht ml). Adjustments usually consist of a gain change that varies linearly with the day since launch of the particular satellite. Two NOAA satellites with AVHRR are usually operating at any given time: one is an afternoon (1330 or 1430 local equator crossing time (ECT)) and the other morning (0730 ECT) Sun-synchronous orbit. Operational calibrations are only performed for the afternoon orbiter. Degradation rates have been computed for NOAA-9 (Loeb 1997); 9, 10, and 11 (Masonis and Warren 2001); and 14 (Tahnk and Coakley 2001) using polar snow targets. For detection of long term trends in remotely sensed parameters and to detect changes in surface or cloud properties over a given location, it is critical to have consistent calibrations over the record of observations from satellite imagers. The polar inter- calibration technique can be used to determine the relative degradation between any two satellites. Thus, given enough overlap, a single calibration reference can be transferred from one satellite pair to the next to provide a consistent calibration for all available AVHRR data. This paper presents preliminary gain comparisons of sequential afternoon and morning satellites from NOAA-9 to NOAA-15 and they are normalized to a well- calibrated reference satellite imager. Recently launched scanning radiometers, including the Moderate Resolution Imaging Spectroradiometer (MODIS), Visible Infrared Scanner (VIRS) and Along Track Scanning Radiometer (ATSR-2) use solar diffusers to calibrate their visible radiances. If these calibrations are reliable and the spectral response functions similar, then one could inter-calibrate VIRS, MODIS and ATSR-2 with AVHRR. It has been shown that VIRS has a stable visible calibration when compared with ATSR-2, MODIS, and other geostationary satellites * Corresponding Author Address: D. R. Doelling, AS&M Inc., 1 Enterprise Parkway, Hampton, VA 23666; e-mail: d.r.doelling@larc.nasa.gov (Minnis et. al. 2001). This paper uses the VIRS data as a reference for the NOAA-14 AVHRR visible channel that will be used to intercalibrate the AVHRRs on the other NOAA satellites. The direct AVHRR-VIRS calibrations are compared with an indirect 3-way calibration using matched VIRS and Geostationary Operational Environmental Satellite (GOES-8) data and AVHRR- GOES calibration. This 3-way calibration provides an estimate of the uncertainty resulting from a given satellite-to-satellite calibration transfer. 2. NOAA Cross-Calibration Method AVHRR Global Area Coverage (GAC) data were used for each satellite. To minimize errors, only coincident, collocated and angle-matched radiances were used to compare visible radiances from different satellite platforms. The best opportunities for meeting such criteria for a pair of afternoon and morning polar orbiters occur over polar regions. Their ground tracks only intersect at 80° north (1000 LST) or south (2200 LST) latitude 14 times a day. During the 8-day satellite repeat cycle, the time difference between the morning and afternoon satellite intersections ranges from 0 up to 45 minutes, the time required to complete half an orbit. Daytime intersections occur in the Northern Hemisphere from February until November; while during the remainder of the year the daytime intersections occur over the Southern Hemisphere. For this study, two days (8 days apart) of intersections are used, when the time difference is less than 2.5 minutes. These days straddle Dec. 23, Apr. 23, June 23, and Aug. 23. The solar zenith angles (SZA), 77°, 67°, 57° and 67°, vary interannually by less than 3° for each month, despite the gradual precession of the ECT for each satellite. For each intersection, several matched data points were created by computing the average 10-bit visible count over selected areas defined by a nominal radius of 50 km from the center of each matched region to minimize the effects of any navigation errors from either one or both of the satellites. These averages are based on approximately 150 pixels from each satellite. Usually, the counts are linearly related to radiance. For the one exception, NOAA-15, the dual-gain counts were first converted to radiances and then converted to NOAA-14 equivalent counts with the nominal calibration to provide a linear fit between counts and gain. Several data points are possible for each intersection because the satellites cross at an angle of