Sundial: A method for inferring image acquisition time from shadow orientation

Abstract Aerial photography and satellite imagery can be used to characterize landscape change over time and help to understand how these changes are related to climate and hydrology. Publicly available optical imagery from sources such as the United States National Agricultural Imagery Program (NAIP) is particularly valuable in this context due to its high temporal and spatial resolution. However, the exact time an image was acquired is often unknown, which complicates, if not precludes, linking images with other types of high temporal resolution data, such as streamflow records. In this letter, we propose a ‘sundial method’ to infer image acquisition time from shadow orientation. This approach involves measuring the direction of a shadow on the image and using solar geometry calculated for the known image date and location to infer the former sun position. Time estimates for 16 Worldview satellite and six NAIP aerial images based on 407 independent measurements of shadow orientation demonstrate the sundial method had an error of 2.1 ± 3.4 min, indicating that image acquisition times can be inferred with a high degree of accuracy and precision. Sensitivity analyses confirm the robustness of the method across different object types, shadow lengths, and solar zenith angles, while also providing practical guidelines regarding the number of measurements required and errors associated with uncertainty in the image date.