Our research utilizing a relational database of matched weather forecasts and observations has been published in the AGU journal Geophysics Research Letters. (link to article)
We evaluated the output from weather forecast models compared to observations at 210 airports across the United States during the November 2019 to March 2020 winter season. We focused on near-surface air temperature errors in the Global Forecast System (GFS) and High- Resolution Rapid Refresh (HRRR) weather models for different times of day and subsets of observed weather conditions. The GFS is 1°C too warm at night and 2°C too cold during the day in conditions with <= 50% and <= 25% cloud cover. The daily high and low temperatures have smaller errors in the HRRR model, which has different algorithms than the GFS model. Model refinement and development efforts would benefit from a focus on accurate representation of the diurnal cycle of temperatures as this basic characteristic of weather can reveal strengths and weaknesses in the model physics.
First author Ronak Patel was an undergraduate research assistant in the Environment Analytics group when he did the analysis. He started graduate school at Cal Tech in September 2021.
Skip to content
