Quality control was carried out on this publication [Version 2.0] of the data. Specifically, we checked the data for outliers on variables which we concluded followed a normal distribution. A z-score calculation was carried out using the following equation:

abs(x_i - mean(x_month)) / stddev(x_month)

where we took the absolute value of the variable [x_i], then subtracted x_i by the mean of x_i values measured in that month [x_month] (using the DATE column), and then divided over the standard deviation of x_i values measured in that same month [stdev_x_month]. Using values only within the same month was to constrain the calculation to a range of values subject to seasonal variation, that would otherwise yield results that could include, for example, temperatures expected during the summer as well as the winter.

These values were then categorised into three groups: z-scores under 3, between 3-6, and above 6. An additional column was then created alongside each parameter denoting whether the z-score for that value was under 3 (denoted as ''1''), between 3 and 6 (denoted as ''2''), or above 6 (denoted as ''3''). This is to give users the option to filter values in the data that have been flagged as potential outliers based on the z-score calculation above.

Due to the harsh conditions in Coal Nunatak, data has not been recorded continually across the period. Missing values, where data was not recorded, are included in the dataset.

The exact locations of sensors within the soil does change over time due to environmental conditions, so depths recorded are a guide.

For columns ''wind speed'' and ''wind direction'', values of 0 may indicate that the sensor was frozen.