This document discusses improved methods for predicting the spectral impact on PV systems based on commonly available data. It presents a modified CREST model that uses the true spectral response rather than a useful fraction to better account for variations due to cloudy conditions. Evaluation shows this modified CREST-WUF model significantly outperforms existing models in predicting daily spectral impact factors for different module types. While this approach captures more variation than previous models, further work is still needed to fully characterize the relationship between site properties and spectral offset to enable general predictions of spectral impact.