Surface roughness is a critical indicator of the health of turbine blades, due to its implications on blade surface heat transfer and structural integrity. The present work proposes a physics-based online assessment framework for industrial gas turbine engines (GTE), in order to assess the blade surface roughness in a high-pressure turbine without engine shutdown. The framework consolidates gas path analysis (GPA) based performance monitoring models and meanline turbomachinery analysis using a novel GPA-meanline matching process. This extracts meaningful performance deviation trends from GPA, while resolving the uncertainties associated with the measurements and modeling. To relate efficiency loss to surface roughness severity, a meanline-based system-identification process has been developed to establish the meanline representation of the turbine stage and to incorporate an empirical surface roughness loss correlation system.