Comparison of Forecasting Biases over New York State Mesonet: A Wet Summer Versus a Dry Summer

Extreme weather events are occurring with increasing frequent due to the climate change. This increasing frequency may introduce more uncertainty in weather forecasting model performance, particularly when considering the intricate relationship of the land surface and atmosphere coupling system. In this study, we utilize data from the sophisticated New York State Mesonet to evaluate the performance of a forecasting system based on WRF Version 4 model, drawing insights from both dry and wet summers. Additionally, the model's performance is assessed on two land surface types: forest and farmland, to provide a comprehensive evaluation of impact of land surface heterogeneity. The surface meteorology, fluxes, and cloud development are assessed. The coupling between surface and atmosphere is diagnosed using a mixing diagram which serves to represent surface thermodynamic properties. The results reveal a systematic increase in warm season dry and warm biases, especially for forested sites during a drought year. The model exhibits heightened sensitivity to drought conditions, resulting in a substantial underestimation of latent heat fluxes during such period. During days with boundary layer clouds, the mixing diagram shows a notably slower growth of moist static energy in the model compared to observation. It is possible that these biases partly attribute to the underestimation of cloud optical depth due to not enough energy for the cloud development. It is essential to understand the weather forecasting model performance under the climate change scenario, particularly as extreme weather events become more frequent. These extreme weather events can increase the level of uncertainty in weather forecasting models and potentially exacerbate model biases. Model forecasts of variables such as soil moisture, surface meteorology and surface fluxes are crucial components of land surface and atmosphere coupling processes. Accurately forecasting of these variables is essential for simulating convection and estimating solar and wind energy production. In our study, we utilize the sophisticated New York State Mesonet observations from both dry and wet summers and discover that these biases increased significantly during dry year. These enhanced biases in dry year may possibly result in the underestimation of boundary layer cumulus clouds. Our forecast system using WRF model underestimates soil moisture in dry year and overestimates it in wet year over farmland The dry and warm biases of 2-m temperature and specific humidity increased significantly over dry year, especially over forest sites Enhanced surface biases in dry year could lead to cumulus cloud underestimation in the model