Bifacial photovoltaic (PV) systems harvest additional light reflected from the ground. The proportion of light reflected from the ground (albedo) is a critical parameter in estimating the expected performance of these systems. It is important that albedo data are available at the system design stage are accurate and relevant, and that the confidence in these data can be quantified. This work investigates the challenges of different albedo data sources and proposes solutions towards more reliable datasets. Satellite sources, such as the NASA Moderate Resolution Imaging Spectrometer (MODIS) products, provide data that can be used to model effective albedo anywhere in the world. Nevertheless, there are numerous sources of error that can add uncertainty in bifacial gain, reducing confidence and increasing financial risk. For smaller sites (<∼10 MW) located in varied landscapes, the scaling error caused by the low spatial resolution can be large. By analysing case studies in complex agricultural landscapes, we investigate the accuracy of different albedo measurement approaches. Monitoring stations with reference cells and pyranometers were deployed to validate satellite data and to realise a site measurement campaign for albedo measurements. Since these sites are much smaller than a MODIS pixel, we have evaluated alternative approaches, including “typical landscapes” with adjustments for local latitude and weather, and a novel method of combining MODIS data with higher resolution satellite data. The latter offers a practical solution with good agreement with the ground-based station.