A number of corrections and adjustments to the raw data were applied for this dataset. For further details see Elvidge et al. (2020).

1) The temperature sensor is not artificially ventilated. This leads to a positive bias in temperature measurements under conditions of high insolation and low wind speed. Simultaneous observations from a thin-wire thermocouple provided a correction procedure, detailed in Smeets et al. (2018).

2) The relative humidity sensor is placed within the unit and experiences elevated temperature under sunny conditions. Internally-observed temperature of the sensor allows the calculation of a dew point, which gives a relative humidity for the observed air temperature. Next, relative humidity is corrected for observations over ice rather than over water. And finally, a well-known underestimation of RH-observations at low temperatures is corrected for by fitting a polynomial through the upper percentile of data and define that polynomial as a RH of 100%. Relative humidity can sometimes greatly exceed 100%, especially in very cold conditions. It is unknown if this is a sensor characteristic or real supersaturation. If required, RH can be capped at 100% without a large error in the corresponding absolute, specific humidity.

3) Longwave radiation observations are corrected for the internal instrument temperature, and for window heating under high insolation conditions.

4) Acoustic snow height observations are corrected for the dependency of sound propagation on air temperature.

5) A few small data 30- to 60-minute data gaps were filled by linear interpolation.