Conventional methods for rock mass classification such as cell mapping and scanline survey have limitation because of human bias and hazard to access rock wall. Development of data collection methods utilizing hyperspectral imagery and terrestrial laser scanning enables engineers to obtain more sophisticated and accurate information about rock mass conditions, without human bias. This paper shows that rock mass parameters can be investigated from those two imagery systems. Hyperspectral imagery identifies weathering and alteration zone, and terrestrial laser scanning indicates orientation, roughness, and joint spacing of a given rock slope. The supervised learning procedure with a small training image is used to understand sitespecific or area-specific discontinuity and weathering trends. The final result shows quantified values for rock mass parameters with a 3D digital geology model fused with a hyperspectral thematic image. The site-specific rock mass representation by the proposed method can be advisable to reduce the time required for the survey in a hazardous environment, and provide consistent classification results regardless of the surveyor.