Calibrating the data logger…
This dataset represents estimates of uncertainty in the day of year (i.e. , "Julian Day") of the onset of the seasonal snowpack. Specifically these are prediction standard deviations for the last day of bare ground derived from long-term time-series of Landsat TM, ETM, and OLI imagery starting in 1993. These maps combine monthly ground snow cover fraction maps from the USGS Landsat Collection 2 Level 3 fSCA Statistics (https://doi.org/10.5066/F7VQ31ZQ), with a time-series analysis of a spectral snow index (NDSI) using a heirarchical Bayesian model (Gao et al. 2021). The combination of these two approaches allows reconstruction of detailed annual snow persistence maps from sparse imagery time-series (Landsat data have an 8 to 16-day return interval in the absence of clouds). A comparison of these data to independent in-situ observations from SNOTEL and microclimate sensors show that these products capture about 85% of spatial variation in snow persistence for recent years (2021-2022), and greater than 90% of temporal variation across the full 1993 - 2022 time-series. This dataset can be combined with prediction means to get approximate prediction intervals. For example, lower and upper 95% prediction intervals could be computed as: lwr_95 = prediction_mean - 1.96 * prediction_sd upr_95 = prediction_mean + 1.96 * prediction_sd These intervals are expected to contain the true value approximately 95% of the time. References: Gao, X., Gray, J. M., & Reich, B. J. (2021). Long-term, medium spatial resolution annual land surface phenology with a Bayesian hierarchical model. Remote Sensing of Environment, 261, 112484. https://doi.org/10.1016/j.rse.2021.112484
Items connected by shared entities, co-authorship, citations, or semantic similarity.