Date of Award

Spring 5-2021

Document Type

Honors Thesis

Degree Name

Bachelor of Science


Atmospheric and Environmental Sciences

Advisor/Committee Chair

Justin Minder


Accurately predicting the type of precipitation in a given winter storm remains a forecast challenge. High-resolution numerical weather prediction models, such as those that make up the operational High-Resolution Ensemble Forecast (HREF) system, are an important tool in the forecast process. However, there is still a large amount of discrepancy between individual members of the forecast ensemble and between the numerical forecasts and observations. These discrepancies can be caused by small differences in the vertical temperature profile leading to large inconsistencies between observational precipitation type (p-type) and the model output. This study will investigate near-freezing precipitation during a winter storm to evaluate the accuracy of the simulated p-type, examine what synoptic or local meteorological features contribute to model errors, and explore how the performance of specific HREF members depends on choices of model physics and initial/boundary forcing conditions. HREF forecasts will be compared to observations from meteorological Phenomena Identification Near the Ground (mPING), the New York State (NYS) Mesonet, and Automated Surface Observing System (ASOS). The winter storm that will be investigated took place on 12-13 January 2018. The large storm entered western NYS the morning of the 12 January, bringing heavy rain, but with a strong cold front moving southeast through the state came a mix of ice pellets, freezing rain, and snow. Even just one day out from the storm, the Buffalo National Weather Service Forecast Office was unsure of the strength and placement of the cold front so the transition areas between precipitation types was uncertain. Initialized at 1200 UTC 12 January, the different HREF ensemble members were inconsistent (about 8 hours from the first observation of mixed precipitation) with the transition areas of rain to ice pellets/freezing rain mainly around 2000- 2300 UTC. Due to this forecast uncertainty in western NY, this area and time will be the main focus of this study. The most accurate member in this time frame was found to be the NAMNEST since the temperatures in western NY in this model were closest to observed. The other models that had cooler temperatures than observed had rain moving through western NY much earlier than observed.

Available for download on Wednesday, December 01, 2021