Major factors affecting the snow simulations by the JULES in New Zealand

Abstract

AbstractSnow frequently occurs over New Zealand. The seasonal snow accumulation can be about 3 m deep in early spring at some locations. Thus, for numerical weather prediction and climate modelling over New Zealand, a reliable, sophisticated snow model that can properly describe the physical processes associated with snow cover is required. In the past, owing to a lack of snow observations, modelling and verification of snow processes was limited over New Zealand. In the present study, the multilayer snow scheme of the Joint UK Land Environment Simulator (JULES)—the land surface model used in New Zealand regional weather and climate models—was used to simulate the snow water equivalent (SWE) and snow density at two sites in the South Island of New Zealand. The model captured seasonal and interannual variability in snow accumulation and melt. However, large errors in the simulated SWE (up to 300 mm) were found during snowmelt periods and in relatively warm winters when a significant amount of liquid water was present in snow. Sensitivity tests showed that errors in the simulated long wave radiation (a negative bias of about 6 W/m2), snow albedo (0.05–0.10) and air temperature for snow occurrence ( T c, about 0.5 K) were the major factors causing large errors in the simulated SWE. In addition, the simulated snow density was lower than observed. The relatively warm and humid maritime climate of New Zealand appears to make snow simulation in New Zealand more sensitive to long wave radiation, snow albedo and Tc than for continents.