Abstract
Extreme freeze events (EFEs) represent a rapid and intense fall of environmental temperature in a constrained area due to low-pressure systems at mid-levels of the atmosphere during winter. The semiarid region of Northwestern Mexico is frequently impacted by EFEs, causing damage to crops, livestock, economy, infrastructure, among other productive sectors. Hence, forecasting and prediction tools to reproduce the phenomenon accurately is crucial to minimize costs and the potential development of an early warning system for this kind of natural phenomena. This study evaluates the Weather Research and Forecasting (WRF) model performance to simulate three relevant freeze events in Northwestern Mexico. Different WRF physics parameterization arrangements were applied, and the results were evaluated using in-situ observations from a local network to measure the model performance through statistical errors. Thus, the aim is to find the key WRF model schemes to reproduce hazardous freezes in Northwestern Mexico. The analyses showed that WRF simulations reproduced the spatial distribution of minimum temperatures during each episode, mainly in the north, northeastern, and close to the steep slopes of the domain. The general model performance shows a negative bias of daily minimum temperatures. The scale of errors was strongly influenced by the temporal resolution (hourly/daily) of in-situ observations. All configurations for short– and long–wave radiation using Dudhia and RRTM schemes provided a better performance in the tested EFEs. A turbulent kinetic energy model was used as the planetary boundary layer scheme proved to enhance WRF model performance.