ML-ready Dataset for Spatio-Temporal Interpolation using SHIFT Hyperspectral Reflectance

The Surface Biology and Geology (SBG) High Frequency Time Series (SHIFT) campaign took place from February to May 2022, with an additional week of data collection in September, as part of NASA's SBG mission. The study focused on a 640-square-mile (1,656-square-kilometer) area in Santa Barbara County, including the nearby Pacific coastal waters. The primary objective of the SHIFT campaign was to gather a dense time series of airborne Visible to ShortWave Infrared (VSWIR) imaging spectroscopy data. This supports NASA's SBG team by: 1) enabling traceability analyses to assess the scientific value of VSWIR revisit data without relying on multispectral proxies, 2) testing algorithms for consistent performance across seasonal timescales and supporting end-to-end workflows, including data distribution to the scientific community, and 3) offering early adoption test cases for SHIFT application users while fostering partnerships with basic and applied science researchers, such as those at the University of California Santa Barbara's Sedgwick Reserve and The Nature Conservancy’s Jack and Laura Dangermond Preserve. In support of the goals above, and in support of the collaboration between machine learning developments and remote sensing science, we leveraged the high-spatial resolution data and high repeat frequency to select a set of 1-square-kilometer sites: Buellton Agriculture, Goleta Wetland, Dangermond Preserve, Sedgwick Reserve, Downtown Santa Barbara, and White Buffalo Land Trust. Data for each site are provided at 5-meter resolution, along with observing geometries and georeferencing information, in portable and easily assessable netCDF4 format. These sites were chosen for their diverse ecological characteristics, offering multiple perspectives on the seasonal evolution of surface hyperspectral reflectance.