Determination of the Angular Sensitivity of a Multispectral Line Scanner from Image data

The DAEDALUS AADS 1268 is a multispectral line scanner with 11 spectral channels and 716 pixels per line. This paper presents a post-flight calibration method to correct the data for dependence of the detector sensitivity on the scanning angle. An area has to be found where only neglegible BRDF effects are expected across the principle plane for zenith angles smaller than the maximum scanning angle. The area does not need to be homogenous, but it must extend over a whole scan line. In our case the runway of the Nuremberg airport was chosen. The pixels of the scan line acquired when crossing the runway at right angles were divided by the respective pixels of the perpendicular overflight after georegistration. The resulting angular sensitivity functions show variations up to 15 % (depending on channel), similar to findings from a laboratory experiment done 3 years earlier. A comparison with laboratory data from a DAEDALUS operated in Australia shows similar results, except for channels 2 and 8. DESCRIPTION OF THE DETECTOR A scan line from the DAEDALUS line scanner AADS 1268 contains 716 pixels for each of the 11 spectral channels. The maximum scan angle is r = 43 to both sides. The scan starts at the right (when looking into direction of flight heading), so pixel number 0 correponds to a scanning direction to the right as seen from the sensor. Each pixel covers an angular range of 2 43 =716 = 0:12 , ground resolution at a flight height of 300 m is about 0.7 m for nadir. ANGULAR SENSITIVITY FUNCTION ASF The angular sensitivity function ASF( r) is the ratio of the measured radiance at the viewing angle r to the measured radiance at nadir ( r = 0 ). The biggest obstacle in determining the ASF is providing a homogenous illumination source. If the ASF is constant and if the radiance reaching the detector is independent of scanning direction r, all pixels will give the same value. The DLR has performed such a test in the laboratory in 1994, pointing the DAEDALUS into an integrating sphere with a diameter of 2 m. For channels 2 to 9 the difference between maximum and minimum measurement was about 5 %, but almost 20 % for channel 1, see fig. 1. However, it remains questionable to what extent these measurements are influenced by inhomogenities of the integrating sphere. There has been no determination of the ASF immediately prior to the 1997 flight campaign over Nuremberg. Because BRDF (bidirectional reflectance distribution function, see e.g. [1]) effects deduced from the data crucially depend on the ASF, we present a method to derive the ASF from our image data. As the instrument was reconditioned since 1994, the ASF as determined by the DLR is significantly different from the 1997 ASF for some channels. In principle, the ASF can simply be determined from a scan line over a spatially homogeneous target. In practice, it is almost impossible to find targets with the required homogenity. Especially urban areas are characterized by a high spatial variance of reflectance. In order to accomodate for this effect, we divided a scanning line that was obtained crossing the runway of the Nuremberg airport by the georegistered data of an overflight along the runway. For a lambertian surface and a constant ASF, the expected result is 1 for all pixels. This procedure is only possible when there is at least one pair of flight tracks perpendicular to each other and the calibration area is seen from both tracks. In case the BRDF of the calibration area chosen is not known, it is furthermore necessary to use scans perpendicular to the sun azimuth, in order to avoid specular or hot spot effects. But even across the principal plane BRDF effects are possible. However, these effects are symmetric with respect to nadir ( r = 0 ) if the calibration area is rotationally symmetric. Symmetric ASF effects cannot be detected by our procedure if the BRDF of the surface is unknown. APPLICATION The method was tested on a set of airborne DAEDALUS data acquired in 1997 over Nuremberg, Germany. The best suited calibration surface in our data is the runway of the airport of Nuremberg for the following reasons: 1.: BRDF effects of the surface (asphalt) across the principle plane in the angular range covered by DAEDALUS (maximum scan angle: 43 ) are small. The sun angle of i = 40 ensures that there are neither effects from a broad hot spot or a broad specular peak that might be expected for a sun positioned in nadir, nor will there be any strong BRDF effects that typically occur for zenith angles larger than 60 . The relative angle of the scan direction to the sun azimuth is 68:5 for the scan to the left and 111:5 This work was supported by the German Research Foundation (DFG) for the scan to the right. So the relative angle is only 21:5 smaller resp. bigger than 90 corresponding to the direction across the principal plane. 2.: The area stretches from the very left of the DAEDALUS scan to the very right, so that almost all pixels can be included in the investigation. 3.: The reflectance profile of the area is quite homogeneous, although deviations of up to 20 % occur. 4.: The width of the runway is about 50 pixels. Thus averaging over the width will dismiss random sensor noise to a large amount. Small scale inhomogenities will also be smoothed after averaging. 5.: Small landmarks on the side of the runway allow a very exact registration of the along-runway scan onto the cross-runway scan. The registration accuracy is estimated to be about one pixel. 6.: A change in reflectance that occured between the cross-runway scan and the along-runway scan is highly unlikely, in contrast to e.g. streets highly frequented by cars. To obtain the ASF, the steps described below were performed. As image data, we did not use the raw data but the reflectance images, because the reflectance images have been processed [2] with MODTRAN to eliminate atmospheric effects . 1. Correct the images for panoramic distortion. Register the along-runway image to the cross-runway image. 2. Average the values over the width of the runway for both images (correct for BRDF if possible). 3. Divide the cross-runway scan by the along-runway scan. The ratio gives the ASF and is shown in fig. 1 for all channels after normalization to nadir (solid line).