Aggregation options allow you to change the underlying aggregation process. To calculate a single aggregated value for a year, data is bundled together from daily or monthly series.

As temperature (and other climate data) follows cyclical annual patterns, missing data from a consecutive block is considered to be more untrustworthy than sporadic data missing throughout the year. This approach allows for a number of records to be missing for a year, so long as not too many consecutive days are missing.

By default, the bundles are groups of 14 days (fortnights) and each bundle requires 70% of the records (10 days of the 14) to be present for the year to be considered reliable enough for an average mean to be calculated.

Some presets (specifically, the cryosphere reference data – sea ice extent and melt) have a lower threshold applied to them because the data has been previously curated; it is considered sufficiently trustworthy that more of it can be missing while not corrupting the results.

If you make changes to these settings and apply them, your settings will take precedence and override any preset specific settings.

Day grouping

Select groups from weekly, fortnightly, monthly, and half-yearly, amongst other options.

Threshold required to form a valid group (% percentage)

This is a percentage of how many records is considered sufficient to form a valid bundle of data.

Apply

Save your changes and apply them to the chart. These settings will persist as you change locations and datasets within the application.

Clear override

This will reset the settings back to their default (14 days at 70% threshold). Only appears after applying your settings.