A global Lagrangian analysis of near-surface temperature extremes: Dataset

Introduction:

This dataset is the basis for the data analysis presented in Mayer (2025). The dataset provides Lagrangian temperature anomaly decompositions (see Mayer and Wirth, 2025) for the near-surface. The dataset was obtained using the tracer method of Mayer and Wirth (2023). It is based on ERA5 reanalysis data (Hersbach et al., 2017).

Data coverage and resolution:

Period: 2010 - 2022

Months: January - December

Temporal resolution: daily

Horizontal resolution: 1° x 1°

Latitudinal extent: -90°N to 90°N

Longitudinal extent: -180°E to 180°E

Vertical extent: Near-surface average, i.e. the vertical average over the lowest 50 hPa of valid values

Applied relaxation constant: 1 / (3 days) 

file type: netcdf

Variables:

The 5 variables contained in the netcdf-files give the individual contributions [in Kelvin] to a given potential temperature anomaly from five different "processes" . 

The netcdf-files contain the variables

"horizontal": contribution from horizontal transport across climatological potential temperature gradients [Kelvin]

"vertical": contribution from vertical transport across climatological potential temperature gradients [Kelvin]

"diabatic": contribution from diabatic heating [Kelvin]

"seasonal": contribution from local changes of the climatological potential temperature including seasonality and the diurnal cycle (small compared to the other contributions) [Kelvin]

"initial": contribution from the pre-existing potential temperature (computed as a residuum of the other 4 terms and the actual potential temperature anomaly) [Kelvin]

with dimensions:

time: time

latitude: latitude [degrees north]

longitude: longitude [degrees east]

The contributions are either given as

absolute contributions (absolutes_mean_50hPa_above_ground.tar),

climatological contributions (climatologies_mean_50hPa_above_ground.tar), or 

anomalous contributions, i.e. contributions relative to their climatological contributions (anomalies_mean_50hPa_above_ground.tar).

Data source and postprocessing:

First, absolute contributions (horizontal, vertical, diabatic, seasonal) were computed using the tracer method of Mayer and Wirth (2023). Output from the tracer method for the individual contributions has been produced for every 3 hours on a global grid with 1° x 1° horizontal resolution and 44 vertical pressure levels. The computation was based on global ERA5 reanalysis data on modellevels (Hersbach et al., 2017; provided by the Copernicus Climate Change Service) with a horizontal resolution of 1° x 1°, on 44 modellevels, and with a temporal resolution of 3 hours. For more information, see README_rawdataproduction.txt.

Second, the contribution from the pre-existing anomaly was computed as a residuum of the other 4 terms and the actual potential temperature anomaly.

Third, all fields were aggregated to yield daily means.

Forth, all variables were combined into one data set.

Climatological means of the daily mean absolute contributions were obtained by first computing the temporal averages specific for each day of the year followed by a  smoothing employing a moving average to the day-specific averages with a window size of +/-15 days.

Anomalous contributions were computed as the difference between the absolute contributions and their day-specific climatological means.

The near-surface averages were computed by averaging the lowest 50 hPa above the surface. The upper most value per grid point assigned to nan has been interpreted as marking the pressure level of the surface (height_suf). The top of the layer to be averaged over (height_top) has been determined as height_top = height_surf + 50 hPa.  Within this pressure range [height_surf, height_top] the values were linearly interpolated to pressure levels at intervals of 5 hPa. Then, the vertical average over these values has been computed.

Compression using the "scale_factor" attribute was applied where still needed and metadata were added.

The code used during the postprocessing is provided in data_processing.zip. For more information, see README_postprocessing.txt.