Wednesday, April 19, 2017, 11:00AM - 12:00PM
Abigail Ahlert and Alexandra Jahn
RL-2 room 155
30th St., CU Boulder East Campus
Climate models in CMIP3 and CMIP5 show a large spread in their simulated Arctic sea ice cover, with large mean biases of sea ice cover in many models. One contributing factor could be that simulated melt season lengths are too short or too long. However, there are multiple possible physical definitions for sea ice melt and freeze onset in climate models, and none of them exactly correspond to the brightness temperature-based melt and freeze onset definitions used for satellite retrievals. This makes the comparison of climate model output to satellite data challenging, as different definitions can bracket the satellite derived melt season dates. To assess the influence of various melt and freeze onset definitions, we compare data from the Community Earth System Model Large Ensemble (available for 1920-2100) to passive microwave-derived melt onset/freeze-up dates (available for 1979-2014). Melt onset definitions for the Large Ensemble are derived from thermodynamic volume tendency, surface temperature and snowmelt, while freeze onset definitions are derived from thermodynamic volume tendency, surface temperature, frazil ice growth, and congelation ice growth. By determining how the modeled melt and freeze onset vary based on different definitions within the model, we gain insight to the interannual, spatial and internal variability of melt and freeze onset dates. This allows us to better assess the level of agreement with the satellite-derived melt and freeze onset dates, and to highlight the challenges in identifying model biases through model-to-satellite comparisons. This analysis will be useful in the upcoming assessment of CMIP6 simulations for melt season biases and can inform sea ice predictability studies that make use of the spring-time melt onset for the predictability of ice advance.
Free and open to the public.