**4. Discussion**

Snow microstructure is an important parameter for microwave and optical remote sensing of snow. Field observations are needed in the development and validation of retrieval algorithms. Therefore, simple and accurate novel measurement methods are required. In this study, the newly developed QST instrument for reflectance measurements was tested on taiga snow, and the empirical relationship between QST reflectance and the microstructural parameter SSA was defined.

We tested the measurement accuracy of SSA to confirm the repeatability of the measurement method. The accuracy was tested by repeating IceCube measurements (several measurements from the same sample) and repeating IceCube sampling (several samples from the same height of snow). The first one resulted in an error (relative standard deviation) of 2.0% and the second one resulted in an error of 2.9%. However, the repeated sampling was made for two of the most difficult types of snow to measure with IceCube with a relatively small set of observations. The total error was below 5%, and measurements are therefore considered repeatable in the taiga snow conditions. We hypothesized that snow sampling for IceCube measurements might change the microstructure of snow. However, the correlation of the two QST reflectance profiles (one directly from the snowpack and one from the sampled snow) was high, with correlation coefficients of 0.92–0.94. Based on this result, we assume that the IceCube sampling procedure does not remarkably affect the optical properties and microstructure of the snow in the sample. Analysis made from the IceCube and the QST measurements resulted in a

strong correlation between SSA and Q, where Q is the ratio of reflectance at the bottom of the absorption feature (1260 nm) and the reflectance change in the absorption feature (between 1160 and 1260 nm). The reflectance profiles measured from both the snowpack and the sampled snow were used to calculate the correlation coefficients, which had values 0.85–0.98. In Section 1, the presented correlations of instruments measuring reflectance or albedo to SSA measurements are around 0.9. Compared to those values, the correlation between SSA and Q is approximately similar or better. Lower values of Q from the sampled snow compared to observations from the snowpack (bias of 0.2) originate likely from absorption by the black sample holder. The outlier values of the sampled snow on 21 March originate most likely from the melting of the snow samples due the longer measurement procedure, since the samples were measured first three times with IceCube, then weighed and finally measured with QST. Typically, the sampled snow was measured with QST directly after one IceCube measurement.

The QST measurements contain error arising from instrument metrics, the observer, and the environmental conditions. As a heavy instrument (2.5 kg), QST is difficult to hold stably in a stationary position with physical contact to the snow when it is not possible to lean the instrument on snow. This is the case with newly fallen surface snow with a low density (around 100 kg m<sup>−</sup>3) and depth hoar layer snow, which is coarse and therefore fragile. This could be avoided by using an appropriate, assembled tripod or another support structure. On the other hand, adjusting the height of the tripod between each measurement of a profile might slow down the measurements. Physical contact to the snow is required so this needs to be confirmed when using external support. However, dense and compact snow (rounded grains or faceted crystals with traditional grain size <1.25 mm and density >200 km m<sup>−</sup>3) had no such problems. In surface snow, solar radiation penetrates the snowpack and possibly causes some degree of error in the measurements. This could be avoided by covering the snowpack, as presented in [7,19]. The sampled snow measurements with QST were made mostly in the shade of the snow pit. The effect of different external illumination conditions was not studied. Warming of the instrument window and plastic casing causes some melting of the snow, and probably some inaccuracy to the measurements, especially during clear skies and positive air temperature conditions. Definition of the exact point of measurement in the snow profile is difficult, with the accuracy being approximately ±2 cm. The instrument size limits its ability to perform measurements close to the ground, so that the bottom of the snowpack below approximately 10 cm is difficult to measure. The clearly erroneous reflectance measurements, which were removed from the analysis, were expected to originate from the wetness of the snow or scattering of the radiation to outside of the field of view.

QST has similar optical geometry to the Contact Probe, and it uses the same range of wavelengths. However, QST is a compact, hand-held instrument while the Contact Probe is attached to a spectrometer. The hand-held instrument has no external spectrometer, laptop, and connecting cables, so QST is therefore fast and simple to use compared to the Contact Probe. In addition, audio notes are possible to record with QST. However, the Contact Probe has less direct contact with snow and the light source is further from the snow, which may reduce additional warming and melting of the snow. Reflectance from the Contact Probe is successfully used for calculation of optical grain size with the Nolin–Dozier model [7]. Therefore, we assume that derivation of optical grain size or SSA from the QST reflectance could be possible with a lookup table in the future, since the empirical relationship between SSA and reflectance has been found. However, it would require a more comprehensive set of observations and proper testing of the penetration depth of radiation and the lost portion of scattered radiation, where field experiments with both QST and the Contact Probe would be beneficial.

**Author Contributions:** Conceptualization, L.L. and A.K.; Formal analysis, L.L.; Investigation, L.L.; Methodology, L.L. and A.K.; Visualization, L.L.; Writing—original draft, L.L.; Writing—review & editing, L.L. and A.K.

**Funding:** Work of Leena Leppänen has been partly undertaken with funding from the Vilho, Yrjö and Kalle Väisälä Foundation of the Finnish Academy of Science and Letters.

**Acknowledgments:** We thank the personnel of Arctic Space Centre of Finnish Meteorological Institute for performing the manual snow measurements. We thank ASD Goetz instrument support program for the availability of the QualitySpec Trek instrument.

**Conflicts of Interest:** The authors declare no conflict of interest.
