In this study, the performance and accuracy of a method for converting airborne multispectral data to reflectance data are characterized. Spectral reflectance is the ratio of reflected to incident radiant flux and it may have values only in the interval 0-1, inclusive. Reflectance is a key physical property of a surface and is empirically derived from on-ground observations. The paper presents a method for processing multispectral data acquired by an unmanned aerial vehicle (UAV) platform, called AggieAirTM, and a process for converting raw digital numbers to calibrated reflectance values. Imagery is acquired by two identical sets of cameras. One set is aboard the UAV and the other is over a barium sulfate reference panel. The cameras have identical settings. The major steps for producing the reflectance data involve the calibration of the reference panel, calibration of the multispectral UAV cameras, zenith angle calculations and image processing. The method converts airborne multispectral data by calculating the ratio of linearly-interpolated reference values from the pre- and post-flight reference panel readings. The flight interval is typically approximately 30 minutes and the imagery is acquired around local solar noon. The UAV is typically flown at low altitudes to reduce atmospheric effects to a negligible level. Data acquired over wetlands near Great Salt Lake, Utah is used to illustrate ground data and processed imagery. The spectral resolution of the multispectral data is 25 cms. The paper discusses the accuracy issues and errors associated with the proposed method.
Available at: http://works.bepress.com/austin-jensen/30/