Datasets of "Machine Learning Reveals the Contrasting Roles of Rainfall and Canopy Metrics on the Formation of Canopy Drip and Splash Throughfall"

This dataset comprises raindrop data and metrics related to canopy structures and applied rainfall collected at the large-scale rainfall simulator at the National Research Institute for Earth Science and Disaster Resilience (NIED) in 2016. Detailed information regarding this study is available in Nanko et al. (accepted, Journal of Geophysical Research: Biogeosciences).

#### Contents

`metrics.zip`

`metrics_treeTrait.csv`: Contains metrics of tree traits.

`metrics_localCrown.csv`: Contains metrics of local crown structures measured at the laser disdrometer.

`raindrop1 – {foliation}_{tree}.zip`

Files include timestamped data from the rainfall simulation, detailing the diameter and velocity of each raindrop measured by the laser disdrometer. Each file is named as: `RD1_EachDrop[{foliation}_{disdrometer location}_{rain rate setting}].csv`.

`raindrop2.zip`

This archive includes aggregated data collected every minute. The metrics provided are the number of drops, Rainfall amount (mm), Rain rate (mm/h), Kinetic energy (J/m²), Normalized kinetic energy by rain amount (J/m²/mm), Drop diameter at specific cumulative drop volume percentiles (0%, 10%, 25%, 50%, 75%, 90%, 100%), Drop count in each 0.1-mm diameter class, and Rainfall amount in each 0.1-mm diameter class.

Files are named as: `RD2_Aggr01min[{foliation}_{disdrometer location}_{rain rate setting}].csv`.

`raindrop5.zip`

This file provides throughfall (TF) partitioning results into splash throughfall (SP), canopy drip (DR), and free throughfall (FR), which are recorded every minute. Included data are: Rainfall amount of OP and TF, Number of drops for OP and TF, D50_OP and DMax_OP, Volume fraction of TF (p),  Weibull parameters, DMax_SP, fSP, fDR, fFR, Rainfall amount for SP, DR, and FR, Drop diameter of DR and SP at specific cumulative drop volume percentiles (0%, 10%, 25%, 50%, 75%, 90%, 100%), Rainfall amount of SP, FR, and DR in each 0.1-mm diameter class.

Files are named as: `RD5_TFpart01min[{foliation}_{disdrometer location}_{rain rate setting}].csv`.

`data_RF.zip`

Contains data and results from random forest modeling.

`dataset_02min.txt`: Source dataset for 2-minute interval modeling.

`relativeImportance.txt`: Variable importance from the random forest model.

`partialDependence.txt`: Partial dependence plots from the model.