The Atmospheric Radiation Measurement program Raman lidar was upgraded in 2004 with a new data system that provides simultaneous measurements of both the photomultiplier analog output voltage and photon counts. We describe recent improvements to the algorithm used to merge these two signals into a single signal with improved dynamic range. The effect of modifications to the algorithm are evaluated by comparing profiles of water vapor mixing ratio from the lidar with radiosonde measurements over a six month period. The modifications that were implemented resulted in a reduction of the mean bias in the daytime water vapor mixing ratio from a 3% dry bias to well within 1%. This improvement was obtained by ignoring the temporal variation of the glue coefficients and using only the nighttime average glue coefficients throughout the entire diurnal cycle.