The dataset is a collection of drone videos collected to calculate surface water flow speeds (velocity) in order to provide field validation of flow speeds calculated from satellite-derived video for the FluViSat project (Hydrological Flow Measurement from Satellite Video). The videos were taken using standard, unmodified DJI-Air2S camera drones on four rivers across England, Australia, Switzerland, and Japan and one tidal site in Scotland. Each video is at least 20-seconds in length, with minimum resolution of 1920 x 1080 pixels. The videos are taken at nadir angle with the drone stationary above a fixed point in the centre of the river. As such a viewer can clearly observe the surface of the water in these videos. These videos were used to provide a very effective validation dataset for the satellite derived videos. Videos were collected on the following dates and when the site was in the following conditions: - Burdekin River, Australia – 7th and 8th January and 18th May. River in high flow conditions. - River Tweed, Norham, England – 15th March. River at medium flows. - Falls of Lora, Connel, Scotland – 16th and 18th March and 14th, 15th, and 16th July. Tidal race captured at various flow conditions. - Uono River, Negoya, Japan – 10th May. River at medium-high flows. - River Rhine at Rhine Falls, Switzerland – 18th August. River at low to medium flows. This work was carried out under the Future-EO-1 EO Science for Society, and funded by, the European Space Agency. Full details about this nonGeographicDataset can be found at https://doi.org/10.5285/d62347ba-0173-4857-b3fc-b7238988f0d8