The fine-scale distribution of snow is important for local water fluxes, avalanche hazards, and refugia. Most models lack the physical processes to represent snow at wind-drift spatial scales, and for those that do, meteorological forcing at these spatial scales are often uncertain. Fortunately, the processes that drive fine-scale snow distribution cause spatial differences in the date of snow disappearance. Here, using a random-forest model to classify snow presence and absence from PlanetScope’s constellation of cubesats, we calculated the date of snow disappearance between 2019 and 2023 for seven alpine meadows in California and Colorado. Snow disappearance date was then used to classify grid cells with annually-repeatable spring snow persistence characteristics. Finally, by combining 1) the date of snow disappearance, 2) snow persistence classes, and 3) in-situ observations of radiation and snow water equivalent (SWE) from nearby snow pillows, we developed a simple approach to reconstruct 3 m and daily spring SWE evolution. Results showed that snow disappearance timing was certain to within: 1 day for approximately 27% of the data (all grid cells in all years), and within 14 days for 95% of the data. Relative to independent lidar data, reconstructed spring SWE had an average spatial coefficient of correlation of 0.74, and mean SWE bias of less than 0.08 m. These results demonstrate that information provided from high-resolution and frequent commercial observations of snow cover could provide a basis calibrating physically based models and facilitating connections between fine-scale snow evolution and coarser scale simulations and observations.