The change of rotor broadband noise due to different surface roughness heights is investigated. The goal of the present work is to quantify the surface roughness at the early stage of helicopter rotor ice accretion through acoustic measurement. Proof-of-concept measurements have been performed in the Adverse Environment Rotor Test Stand (AERTS) facility at The Pennsylvania State University to explore the effect of different surface roughness heights on rotor broadband noise. First, sand paper with different grit sizes is applied to the leading edge of the blade to represent different surface roughness at different ice accretion times. The measured broadband noise, in the frequency bandwidth region ranging from 10 kHz to 24 kHz, is separated by at least 2dB due to due to different sand paper grit sizes. This demonstrates the feasibility of quantifying surface roughness through acoustic measurements. Comprehensive broadband noise measurements based on different accreted ice roughness are then conducted to form the data base from which a correlation between the ice-induced surface roughness and the broadband noise level is developed. Two parameters, the arithmetic average roughness height, Ra, and the averaged roughness height, based on the integrated ice thickness at the blade tip, are introduced to describe the ice-induced surface roughness at the early stage of the ice accretion. The ice roughness measurements are correlated to the measured broadband noise level. Strong correlations (absolute mean deviations of 9.3% and 11.2% for correlation using Ra and the averaged roughness height separately) between the ice roughness and the broadband noise level are obtained, which can be used as an early ice accretion detection tool for helicopters, as well as to quantify the ice-induced roughness at the early stage of rotor ice accretion.