WRF model performance analysis for a suite of simulation design

At present scientists are successfully using Numerical Weather Prediction (NWP) models to achieve a reliable forecast. Nested domains are preferred by the modelling community with varying grid ratios having wider applications. The impact of the nesting grid ratio (NGR) on the model performance needs systematic analysis and explored in the present study. The usage of WRF is mostly as a mesoscale model in simulating either extreme events or events of smaller duration shown with statistical model evaluation for the correspondingly similar and short period of time. Thus, influence of the simulation period on model performance has been examined for key meteorological parameters. Several works done earlier on episodes involve model implementation for longer duration and for that single simulation is performed often for a continuous stretch. This study scrutinizes the influence on model performance due to one single simulation versus several smaller simulations for the same duration; essentially splitting the run-time. In the present study, the surface wind (i.e., winds at 10 meters), temperature and Relative humidity at 2 meters as obtained from model simulations are compared with the Observations. The sensitivity study of nesting grid ratio, continuous versus smaller split simulations and realistic simulation period is done in the present study. It is found that there is no statistically significant difference in the simulated results on changing the nesting grid ratio while the smaller time split schemes (2days and 4days schemes on comparison with 8days and 16days continuous run) improve the results significantly. The impact of increasing number of observations from different sites on model performance is also scrutinised. Furthermore, conceptual framework is provided for Optimum time period for simulations to have confidence in statistical model evaluation.