Ocean color remote-sensing systems require highly accurate calibration (<0.5%) for accurate retrieval of water properties. This accuracy is typically achieved by vicarious calibration which is done by comparing the atmospherically corrected remote-sensing data to accurate estimates of the water-leaving radiance. Here we present a new method for vicarious calibration of a hyperspectral sensor that exploits shadows cast by trees and cliffs along coastlines. Hyperspectral Ocean PHILLS imagery was acquired over East Sound and adjacent waters around Orcas Island, Washington, USA, in August, 1998, in concert with field data collection. To vicariously calibrate the PHILLS data, a method was developed employing pixel pairs in tree-shaded and adjacent unshadowed waters, which utilizes the sky radiance dominating the shaded pixel as a known calibration target. Transects extracted from East Sound imagery were calibrated and validated with field data (RMSE = 0.00033 sr-1), providing validation of this approach for acquiring calibration-adjustment data from the image itself.