MASCDB: a database of images, descriptors and1microphysical properties of individual snowflakes in free fall

Dataset overview

This dataset provides data and images of snowflakes in free fall collected with a Multi-Angle Snowflake Camera (MASC) The dataset includes, for each recorded snowflakes:

1. A triplet of gray-scale images corresponding to the three cameras of the MASC
2. A large quantity of geometrical, textural descriptors and the pre-compiled output of published retrieval algorithms as well as basic environmental information at the location and time of each measurement.

The pre-computed descriptors and retrievals are available either individually for each camera view or, some of them, available as descriptors of the triplet as a whole. A non exhaustive list of precomputed quantities includes for example:

• Textural and geometrical descriptors as in Praz et al 2017
• Hydrometeor classification, riming degree estimation, melting identification, as in Praz et al 2017
• Blowing snow identification, as in  Schaer et al 2020
• Mass, volume, gyration estimation, as in Leinonen et al 2021

Data format and structure

The dataset is divided into four .parquet file (for scalar descriptors) and a Zarr database (for the images). A detailed description of the data content and of the data records is available here.

Supporting code

A python-based API is available to manipulate, display and organize the data of our dataset. It can be found on GitHub. See also the code documentation on ReadTheDocs.

Download notes

• All files available here for download should be stored in the same folder, if the python-based API is used
• MASCdb.zarr.zip must be unzipped after download