Browsing by Author "Carver, Cameron"

Now showing 1 - 1 of 1
  • No Thumbnail Available
    Item
    Open Access
    Development of sea ice diagnostic tools for high-resolution simulations of the Climate Model Intercomparison Project (HiResMIP)
    (2025) Carver, Cameron; Vichi, Marcello; Iovino, Dorotea
    Derived from the necessity to more thoroughly understand the role that horizontal resolution plays in the performance of climate modelling, this minor-dissertation describes the development and initial testing of a High-Resolution Sea Ice Diagnostics Toolset. This is designed to evaluate the influence increased horizontal resolution has on the ability of high-resolution climate models to recreate Antarctic sea ice behaviour. Analysis is conducted through the calculation and visualisation of a defined univariate performance metric, referencing a single user defined satellite-derived observational dataset. Developed in Python, cloud-ready datasets and computing capabilities are utilized to eliminate the need for local download of large model datasets and extensive computational capacity. Temporal mean metric values are produced and visualized, providing easy visual analysis of regional model performance for each respective temporal grouping. Spatial mean metric values for each month produce a time series that reveal performance trends. Distributions of these values give insight into the model performance for particular seasons and months. Sensitivity analysis functionality enables the assessment of model performance for specified ranges of observed sea ice concentration values. A preliminary assessment is conducted to substantiate the value of the toolset products. Results indicate alignment with findings of Selivanova et al. (2024) where only marginal improvement in model performance is seen with increased resolution for the models assessed. Additionally, sensitivity analysis results highlight the shortcomings of all assessed models in recreating Antarctic sea ice behaviour across the marginal ice zone. Integration of cloud-ready datasets and cloud computing sees significant reduction in time and increased functionality. While the scope of this minor dissertation limits the depth of analysis, the toolset developed here is shown to provide useful insights into the role of horizontal resolution and establishes backbone from which users can generate and visualise univariate metrics for high resolution datasets with minimal local computational load and significantly reduced runtime.