PREDICTION OF GROUNDWATER LEVEL FLUCTUATIONS ON GROHOVO LANDSLIDE USING RULE BASED REGRESSION

摘要

In order to contribute to understanding the effect of atmospheric conditions on the groundwater level fluctuations on Grohovo landslide, a machine learning tool for induction of models inform of the set of rules was applied on a dataset comprising daily atmospheric and groundwater level data measured in 2012. The atmospheric data comprises of an average daily air temperature, humidity, wind speed, pressure, total evapotranspiration, and precipitations. For the experiment independent variables i.e. atmospheric data and present groundwater level were used to model target variable i.e. predicted groundwater level for 24 and 48 hours in advance. The presented models give predictions 24 (first model) and 48 (second model) hours in advance for groundwater level fluctuations on Grohovo landslide. The first model is consisted from seven, and the second model from five rules. Both models have very high correlation coefficients of 0.99 and 0.97, respectively. From the given models, it can be concluded that the most influence on the groundwater level fluctuations have sum of daily precipitations and average daily air temperature. The obtained models are intended for use in the models for debris flow propagation on the Rjecina River as a part of an Early Warning System.