Fig. 1: Performance metrics for each benchmark phase partitioning method (PPM).

These metrics include accuracy (a), snow bias (b), and rain bias (c) plotted by air temperature for the crowdsourced and synoptic datasets. The vertical dashed line in each panel represents the 0 °C isotherm. Note: the −100% snow and rain bias values in (b) and (c), respectively, correspond to air temperatures where the method does not predict the given phase. For example, Ta1.0 predicts only rain above 1.0°C and thus presents a −100% bias for all observed snowfall above that threshold. Source data are provided as a Source Data file.