Non-hydrostatic hybrid-coordinate modelling: Simulation of extreme weather event on 20–22 April 2005 in Cairo, Egypt

We demonstrate in this study that by modifying a hydrostatic numerical weather prediction (Eta) model to incorporate some non-hydrostatic processes leads to significant improvement in the simulation of extreme weather (sand storm and rainfall) event that occurred over Cairo, Egypt during 20–22 April 2005. We comparesimulations of the event by both the original version of the hydrostatic (HY) and modified version (hereafter non-hydrostatic (NH)) models to evaluate any improvement in skill due to incorporation of non-hydrostatic processes in the latter (NH). The computed skill score of sand storms as simulated by both (NH) and (HY) models for our case show that NH model correctly captures 278 cases from the total 539 events forecast, and falsely captures 261 events compared to 108 and 431 cases for HY, respectively. The NH model reaches perfect skill for the Probability of Detection (POD) and the Detection Failure Ratio (DFR) by the second day and third day of integration. Critical Success Index (CSI) indicates no skill for HY forecast after first day, also no skill of (POD), but has perfect skill of False Alarm Ratio (FAR). True Skill Statistic (TSS) and Heidke reach its highest values at second day of NH and lowest values at third day for HY. model reproduces fairly well the correct non-rainy day events although thethe false alarm rate increases after the third day of model integration. However, the NH model captures the correct number of rainy-day events as well as the episodic events. infall simulated (forecast) by NH and HY models have Probability of Detection (POD) approximately 65% and less than 20%, respectively. The False Alarm Ratio (FAR) by the NH model approaches 87% on second and third days during the 72-hour integration period, while it is 92% and 85% for the HY model by the second and third day, respectively.