Inter-Annual Temporal Repeatability of Snowmelt Patterns Derived from Remote Sensing

Seasonal snowmelt has been noted in the literature for many years to follow a repeatable pattern year after year. Woodruff & Qualls (2019) were the first to develop a method to quantify the repeatable pattern using multiple years of remotely sensed MODIS snow covered area images. In that paper, they demonstrated how well the inter-annual snowmelt pattern for a watershed repeated itself spatially. In this presentation, we focus on the temporal repeatability across years of the melt pattern extracted from the Woodruff & Qualls' (2019) pattern synthesis method. We show that while the timing of melt may shift or translate its window of occurrence and the duration of melt may vary from year to year, the pattern of melt follows the same relative timing from year to year. Accounting for lateral shifts in the timing of melt due to secondary snowfall events, that is, those occurring subsequent to the onset of spring melt, is important for quantifying the recurring temporal pattern of melt. The temporal pattern of melt in individual years compared to the interannually recurring pattern has an RMSE ranging from 2-4% and an R-squared greater than 0.98 from 2000 - 2017. This temporal repeatability is important since it may allow the opportunity to forecast the duration of spring melt across a watershed, once the initial rate of melt has been established.

Woodruff, C. D., and R.J. Qualls, Recurrent snowmelt pattern synthesis using Principal Component Analysis of multi-year remotely sensed snow cover, Water Resources Research, 55(8): 6869-6885, https://doi.org/10.1029/2018WR024546, 2019.