EXAMINATION OF INPUT DATA FOR PREDICTION OF DAM INFLOW DURING SNOWMELT SEASON USING RANDOM FOREST

Abstract

In predicting the inflow of dams during the snowmelt season by deep learning, it's important to select input data. In this study, we visualized the importance of input data using the random forest method. The input data includes precipitation, global solar radiation, reflected solar radiation, upward radiation, downward radiation, surface temperature, temperature, wind speed, humidity, snow weight, snow depth and snowmelt amount.

As a result, the most important factors were the temporal distribution of precipitation, upward radiation, air temperature, surface temperature, and snow depth.