A song of ice and fire: Spatiotemporal visual variations in the snow water equivalent (SWE) for identification of wildfire risk areas in a snow mountainous region

Abstract

Farmers and ranchers living in snow mountainous regions heavily rely on melting snow for agricultural irrigation. However, major natural disturbances, such as wildfires, have markedly reduced melt durations, disrupting the local hydrological cycle. In this study, we estimate the impacts of wildfires on snow accumulation and retention in the San Joaquin watershed of Sierra Nevada around California during the 2000 to 2016 water year through spatiotemporal visual variations in the snow water equivalent (SWE). To better explore the relationship between SWE and wildfire, machine learning methods, including regression analysis and deep analytics, were performed to examine the original remote sensing dataset of SWE in the San Joaquin watershed. The results suggested that wildfires have greatly increased the speed of snowmelt and shortened the timing of water availability, which may result in water resource shortages. We also comparatively analyzed the peak SWE values in each year, in which two water year groups (group (1): 2000–2003, group (2): 2013–2016) were selected and compared based on the frequency of major wildfires. Interestingly, higher fluctuations in SWE curves were observed from 2010 to 2016 than from 2000 to 2003, which suggested that the frequency of wildfires was much higher in recent years than in the past twenty years. Ultimately, the correlation between fire severity and SWE values was determined to be strongly positive (p < 0.05). Overall, our study provides visual-mathematical perspectives for local ecology management authorities to understand the most related factors behind the shortened melt duration.

Keywords

• Snow melting
• Machine learning
• Environmental disasters
• Wildfire
• Snow-water equivalent