In this paper, we propose a new method to retrieve the relationship between the cloud fraction (CF) and the integrated incident surface solar flux (0.3-3um) by considering the multi-reflection effect between cloud ceiling and the snow/ice surface. Using in situ observations from the Canadian High Arctic we examine the sensitivity of each parameter in this new relationship using a range of observed melt onset conditions. Numerical sensitivity tests and observed data show that the derived parabolic curve can fit the observed data with a higher correlation coefficient (0.86) than other contemporary methods. Snowmelt processes are then examined using the CF-SCF relationship. Results show that, when the daily averaged net surface flux approaches 40 W•m-2, the snow cover above a first year landfast sea ice begins to melt. However, the initiation moments of snowmelt are significantly advanced and delayed correspondingly, and the melting periods are shortened and elongated correspondingly, with the assumption of different cloud fractions. Analysis of the efficiency of the SCF shows that for selected cloudy days SCF is highly variable diurnally due to the interplay of solar zenith angle with surface insolation. We conclude the paper with an examination of how the SC-SCF relationship can be formulated using optical and thermal infrared Earth Observation data from MODIS and related satellite and how these data may be assimilated into a thermodynamic snow/sea ice model.