615 Multivariate Ensemble Sensitivity for Typhoon Haiyan

Tuesday, 9 January 2018
Exhibit Hall 3 (ACC) (Austin, Texas)
Sijing Ren, Nanjing Univ., Nanjing, China; and L. Lei

Handout (707.4 kB)

Multivariate ensemble sensitivity has proven a better solution than frequently-used univariate ensemble sensitivity which ignores correlation cross variables and results in overestimations, especially for mesoscale problems. To demonstrate that in a real case, this study chooses Typhoon Haiyan (2013) which shows great challenges in intensity forecast. Based on ensemble forecast of 80 members using the Weather Research and Forecasting (WRF) model, univariate and multivariate ensemble sensitivity are both used to provide estimations of sensitivity, followed by perturbed initial condition experiments. According to results, univariate and multivariate sensitivity are consistent in indicating that increasing intensity is associated with a moistier deep convection region in mid-lower troposphere, a stronger warm core, an increased primary circulation particularly at maximum wind radius and an increased secondary circulation. However, multivariate sensitivity is more accurate in perturbation response prediction, especially for outflow layers of Haiyan.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner