Water level prediction method, water level prediction program and water level prediction device

摘要

PROBLEM TO BE SOLVED: To improve the water level prediction accuracy while reducing the amount of input data when predicting the water level of a river by using machine learning. SOLUTION: In the present disclosure, the rainfall data of "the entire basin" in the upstream area of the water level prediction point is not input, but the rainfall data of "on the river" in the upstream area of the water level prediction point is input. Then, for the rainfall data "on the river" in the upstream area of the water level prediction point, the "convolution process" is not executed, but the "fully combined process only" is executed. Here, the following data (1) or (2) are input as the rainfall data "on the river" in the upstream area of the water level prediction point: (1) "Observation station rainfall" measured at the ground rainfall station. Data, (2) Data of "analyzed rainfall" obtained by correcting the data of radar rainfall measured by radar reflection in the sky using the data of observation station rainfall measured at the rainfall station on the ground. [Selection diagram] FIG. 6