Sensitivity Analysis of the WRF Model to Simulate Precipitation in the Metropolitan Area of the Valley of Mexico for the Period June-September 2019

Abstract

Abstract Every rainy season, it is common for the Metropolitan Zone of the Valley of Mexico to be affected by extreme weather. The objective of this work was to carry out a sensitivity analysis combining the different schemes of physical processes incorporated in the WRF and find the optimal scheme to carry out the weather forecasting of rainfall at high spatial resolution in a densely urbanized zone. For this purpose, rainfall during four months in 2019 (jun-sep), were simulated with 40 different combinations of physical parameters. The performance of the model was evaluated through the Efficiency Multiparametric Index, considering as complementary statistical metrics Bias Percentage, Mean Absolute Error, Mean Square Error, Nash-Sutcliffe Index, and Pearson's Correlation. The results of the model were compared with the observation records, considering five thresholds of rainfall. The sensitivity analysis suggests that there are specific WRF configurations that can predict most rainfall features in the study area, such as the use of Milbrandt Yau Double Moment microphysics, MYNN Level 2.5 planetary boundary layer, and Betts Miller Janjic as cumulus parameterization, this for precipitation thresholds greater than 20 mm. Finally, the applied methodology made it possible to identify the experiment with the best performance and consistency in the different analysis thresholds for the metropolitan zone.

Keywords

• Numerical weather prediction
• Valley of Mexico Metropolitan area
• urban rainfall
• cumulus parameterization
• microphysics parameterization
• planetary boundary layer parameterization