| Summary: | Satellites and several specially equipped scientific aircraft provide basin-wide altimetric measurements of sea ice freeboard, from which sea ice thickness can be estimated. Up to now, few methods have been developed to use these measurements to quantitatively assess the skill of predictive models of sea ice for the Arctic. This thesis addresses this problem, using measured freeboard from ICESat and Operation IceBridge (OIB). Output from the Regional Arctic System Model (RASM) is used to demonstrate applicability of both variance- and correlation-weighted skill scores of freeboard that quantify model skill and take measurement error into account. The techniques are demonstrated using two different RASM configurations, one using Elastic-Viscous-Plastic (EVP) ice mechanics, the other using the Elastic- Anisotropic-Plastic (EAP) rheology, both simulated for 2004 and 2007, during which ICESat was in operation. RASM variance skill scores ranged from 0.712 to 0.824 and correlation skill scores were between 0.319 and 0.511. The skill scores were calculated for monthly periods and require little adaption to be applicable for monthly to decadal Navy forecasts of the Arctic. This will help improve sea ice prediction by quantifying model limitations and thus maximize the usefulness of ICESat-2 freeboard measurements after that satellite is launched next year. Approved for public release; distribution is unlimited. Lieutenant, United States Navy http://archive.org/details/quantifyingseaso1094549374
|
|---|