Assessment of the Experimental Rain Garden Infiltration Rate Using Artificial Intelligence Techniques

This study conducted field experiments on four rain gardens to obtain infiltration rates for various plants planted at various spacing. Two rain gardens were size 2 m × 1 m × 0.15 m, and the others were size 1 m × 1 m × 0.05 m. The experimental data were used to model the infiltration characteristics of rain gardens by four artificial intelligence techniques, i.e., artificial neural networks, distributed random forests, generalized linear models, and random forest. Input parameters for the modeling process encompass air temperature, soil water content, water inflow rate to a rain garden, number of plants, plant height, and time elapsed during different water level drops of accumulated water in a rain garden. The infiltration rate was used as an output parameter for modeling, with 75% of the data allocated for training and the remainder for validation. The performance of artificial intelligence methods was evaluated using statistical metrics: correlation coefficient (CC) and root mean square error (RMSE). Notably, the random forest model demonstrated superior performance compared to other models. For the training dataset, the CC and RMSE values were 0.994 and 0.231 cm/h, respectively. In the validation dataset, these values were 0.894 and 0.798 cm/h. For distributed random forests, the validation dataset, the values 0.783 and 0.837 cm/h. For artificial neural networks, the validation dataset, the values 0.78 and 1.24 cm/h, and for generalized linear models, 0.481 and 1.288 cm/h, respectively. This study’s findings are valuable for accurately evaluating the infiltration rate of the rain garden.