Cloud base height (CBH) is a critical input to short-term solar forecasting algorithms, yet CBH measurements are difficult to obtain. Existing methods to detect CBH include radiosondes, ceilometers, and the stereographic method. However, these methods are deficient for intra-hour forecasting due to high costs or low temporal resolution. While satellite images could overcome these limitations, only the cloud top height can be determined from the thermal IR channel. We describe the integration of a cloud shadow speed sensor (CSS) with angular cloud speed from a sky imager to determine CBH. Furthermore, an improved methodology to determine cloud motion vectors from the CSS is presented, which offers lower noise and greater accuracy and stability than existing methods. Two months at the UC San Diego campus were used for validation against measurements from meteorological aerodrome reports (METAR) and an on-site ceilometer. Typical daily root mean square differences (RMSD) are 126 m which corresponds to 16.9% of the observed CBH. Normalized RMSD remains below 30% for all days. The daily bias is usually less than 80 m which suggests that the method is robust and that most of the RMSD is driven by short-term random fluctuations in CBH. Unlike sky image stereography the present method can be applied to measurements at a single site making it widely applicable.