Approximating snow surface temperature from standard temperature and humidity data: New possibilities for snow model and remote sensing evaluation

摘要

Snow surface temperature (T_s) is important to the snowmelt energy balance and land-atmosphere interactions, but in situ measurements are rare, thus limiting evaluation of remote sensing data sets and distributed models. Here we test simple T_s approximations with standard height (2-4 m) air temperature (T_a), wet-bulb temperature (T_w), and dew point temperature (T_d), which are more readily available than T_s. We used hourly measurements from seven sites to understand which T_s approximation is most robust and how T_s representation varies with climate, time of day, and atmospheric conditions (stability and radiation). T_d approximated T_s with the lowest overall bias, ranging from -2.3 to +2.6℃ across sites and from -2.8 to 1.5℃ across the diurnal cycle. Prior studies have approximated T_s with T-a, which was the least robust predictor of T_s at all sites. Approximation of T_s with T_d was most reliable at night, at sites with infrequent clear sky conditions, and at windier sites (i.e., more frequent turbulent instability). We illustrate how mean daily T_d can help detect surface energy balance bias in a physically based snowmelt model. The results imply that spatial T_d data sets may be useful for evaluating snow models and remote sensing products in data sparse regions, such as alpine, cold prairie, or Arctic regions. To realize this potential, more routine observations of humidity are needed. Improved understanding of T_d variations will advance understanding of T_s in space and time, providing a simple yet robust measure of snow surface feedback to the atmosphere.