Empirical prediction of snow and land-fast sea ice temperature from surface meteorology on Pistolet Bay, Northern Newfoundland

Abstract

The temperature of sea ice and snow cover on the ice surface is a critical component of thermodynamic models of sea ice. Ice temperature is also an important factor in the sea ice strength and therefore in the potential loads the ice can exert on ships, ports, bridges, and offshore structures. The snow layer, while remaining distinct from the sea ice in its physical properties, can become compacted on the ice and contribute to the effective ice thickness and strength. The temperature of sea ice is governed by the atmospheric energy flux balance at the snow surface and the ocean energy flux at the ice base. In addition, snow cover acts as an insulator to sea ice, buffering it against temperature changes due to changes in surface meteorological conditions. The average internal ice temperature and total thickness (through its impact on ice salinity and brine volume) significantly affect ice strength parameters used in the calculation of loads on engineered structures. In February 2017–2019, three field campaigns were carried out on the land-fast ice on Pistolet Bay, north of St. Anthony, Newfoundland in Atlantic Canada. For approximately five days in 2017 and about four days in 2018–2019, surface air temperatures, wind speeds, and snow and ice temperature profiles were continuously measured. Dew point temperatures were obtained from the North American Regional Reanalysis (NARR) dataset. The datasets from the three field campaigns were used to develop empirical linear regression models of snow and ice surface and depth-mean temperatures as a function of the air temperature alone, and of the air and dew point temperatures and wind speed. In this paper, the meteorological data from the three field campaigns are presented, along with the snow and ice thickness and temperature data. The linear models of snow and ice temperature are presented, tested using the observed field data from the Pistolet Bay campaigns, and compared with results obtained from previously published models of snow and sea ice temperature. The results show that the models of snow and ice temperature as a function of the three meteorological variables are modestly more robust than the models as a function of air temperature alone, with R2 values of up to 0.90. The models presented in this paper have potential usefulness in improved prediction of site-specific snow/ice temperatures and associated temperature-dependent ice properties, such as strength at sites where gridded weather data or in-situ meteorological measurements are available, but in-situ snow/ice temperature measurements are nonexistent. Keywords: Northern Newfoundland, Land-fast sea ice, Snow on sea ice, Thermodynamic snow/ice modelling