Almost everything on earth benefits from the services rendered by snow. Snow directly influences the balance between the energy absorbed by the Earth and the energy reflected back into the atmosphere due to its high radiation reflection ability. About 80% to 90% of the total sunlight hitting its surface is reflected back into the atmosphere, thereby cooling off the planet. Above that, snow is the primary water source for many parts of the world. About one-sixth of the world depends on water from snowmelt to support life. Snow is often located in remote regions, and we often rely on remote sensing technologies to measure snow properties.
Understanding snow properties is challenging because, on the one hand, the historical pattern of snow accumulation and snowmelt has been altered due to global warming. On the other hand, remote sensing technologies do not always provide accurate estimates of snow properties due to the high variability in the snow both spatially and temporally. The need for precise prediction motivates the use of Machine Learning techniques for understanding snow properties. In this study, we survey articles that have applied Machine Learning techniques to predict Fractional Snow Cover, Snow Depth, and Snow Water Equivalent. The result of our survey indicates that Neural Networks, Random Forest, and Support Vector Machines are the most used techniques for estimating snow properties. Also, current gaps in the application of Machine Learning to snow science are discussed, and some open questions are presented.