“What’s the Temperature Tomorrow? Increasing Trends in Extreme Volatility of Daily Maximum Temperature in Central and Eastern United States (1950–2019)“, is an analysis released on the Science Direct website this December 2022. The analysis focuses on the behavior of daily maximum temperature volatility in the United States. Volatility, in this analysis, refers to “magnitude of temperature change between two consecutive days”. The aspects of volatility include climatology, seasonality, and directionality (increase vs. decrease).

Mohammed Ombadi and Mark D. Risser, authors of the analysis, discovered an increasing trend of roughly 0.2–0.3 °C/decade in the magnitude of volatility within the Central United States. However, this statistically significant trend was not supported by all the datasets that were analyzed.

Research Method and Design

The daily climate records used in the analysis contained observations from the years 1950- 2019. Only years with less than 5% missing data were taken into consideration to ensure that observations accurately portrayed seasonal variability for each year.

The authors of the study used five datasets in this analysis: GHCNd, GHCN-homogenized, Livneh daily CONUS near-surface gridded meteorological dataset, ERA5, and NOAA20CR. The 1,484 ground station observations from the Global historical climatology network daily (GHCNd). The GHCNd is a mixed database of daily climate summaries recorded and taken from land surface stations around the world. The so-called GHCN-homogenized is a “homogenized” version of the GHCNd dataset. The GHCN-homogenized data accounts for all possible non-climatic sources of data inconsistency, such as station location differences or observation practices. There were 1,839 stations in the GHCN-homogenized dataset. The third dataset used in this analysis is the Livneh daily CONUS near-surface gridded meteorological dataset, which was compared against the outcomes taken from GHCNd. Both the ERA5 and NOAA20CR combine data numerical weather prediction and observations by using data assimilation systems. The ERA5 and NOAA20CR also have greater spatial and temporal extents than GHCNd.

Conclusion Drawing

According to the analysis’ concluding remarks, “results from trend analysis indicate that most stations (>50%) in climate divisions of Upper Midwest, South and Southwestern exhibit increasing trends in  Vmax‾ (the 95th percentile of absolute difference in temperature between two consecutive days) ranging from 0.2 to 0.3 °C.” However, ERA5 and NOAA 20CR show much smaller values for daily maximum temperature volatilities by 25% and 49% compared to GHCN datasets. In other words, the two reanalysis datasets failed to reproduce the observed trends.

The appearance of “extreme volatilities” was determined by the time of year and region being observed. For example, extreme volatilities primarily occur during April, May, and June in the Western United States. During December, January, and February, volatilities peaked on the east side of the Rocky Mountains. In the Great Lakes region, March, April, and May spring months show the most extreme volatilities.