Fig. 7: Overview schematic of our project workflow.

Here, we show the datasets (blue boxes), the tuning, fitting, and validation steps (pink boxes), and the phase partitioning methods (PPMs, peach boxes). We start with complete datasets (a), which include observations of rain (R), snow (S), and mixed precipitation (M) along with air (Ta), wet bulb (Tw), and dew point (Td) temperature, relative humidity (RH), and pressure (P). We split these datasets into training (b) and testing (c). We tune the hyperparameters (d) and fit (e) the random forest (RF), XGBoost (XG), and artificial neural network (ANN) machine learning models (f) using the training data. We then apply the machine learning models and the benchmarks (g, Table 1) to the testing data to validate their precipitation phase predictions (h).