Comparing Simulated, Synthetic Tropical Cyclones in Unstructured Climate Models Downscaled from 25 to 3 Kilometer Grid Spacing
Open Access
- Author:
- De Ciampa, Corrine
- Graduate Program:
- Meteorology and Atmospheric Science
- Degree:
- Master of Science
- Document Type:
- Master Thesis
- Date of Defense:
- June 28, 2023
- Committee Members:
- Colin M. Zarzycki, Thesis Advisor/Co-Advisor
Chris E Forest, Committee Member
Xingchao Chen, Committee Member
Paul Markowski, Program Head/Chair - Keywords:
- tropical cyclones
unstructured global climate models
CAM
MPAS
precipitation extrema - Abstract:
- Tropical cyclones (TCs) can carry the risk of potentially devastating flooding from extreme precipitation - either through intense bursts of heavy precipitation, prolonged, constant precipitation, or a combination of both. Realistically replicating TCs in climate model simulations is necessary to further understand these high-impact scenarios. Previous studies mainly rely on simulating TCs in global climate models (GCMs) with 28 km grid spacing, though 3 km grid spacing is required to adequately simulate realistic TC features and small-scale dynamical processes. To understand the influence of model resolution on TC precipitation extrema and the potential hydrologic value-added, we filter intense, synthetic TCs making landfall in South Florida within an existing, ~300-year, 28 km climate model ensemble generated by the Community Atmosphere Model, version 5 (CAM5). The atmospheric states from this subset of TCs are used to generate dynamically downscaled versions of these same synthetic TCs at 3 km grid spacing by coupling CAM5 to the dynamical core of the Model for Prediction Across Scales (MPAS). Both models are derivatives of CAM and rely on CAMv5 physics to isolate the impact of model resolution on storm evolution. The 28 km "parent" simulations are compared to the 3 km "child" simulations to evaluate changes in TC structure, intensity, and precipitation extrema, with an emphasis on key metrics relevant to our stakeholders in South Florida. The 3 km child simulations outperform the 28 km parent simulations in qualitative storm structure, including realistic-looking eyewall and outer rainband features, whereas these features are poorly replicated in the 28 km parent simulations. Additionally, the 3 km child simulations indicate heavier precipitation over shorter time periods and also more prolonged precipitation overall than their 28 km parent counterparts. However, the value added by 3 km grid spacing with respect to storm total precipitation remains unclear. These findings suggest that this framework could be used for further investigation through ensemble validation.