Soil Moisture Prediction Using Hydrological Model And Machine Learning Method

Chu Kang, Chien-Chih Chen, Ming-Hsu Li, Tsung-Hsi Wu, Meng-Ju Chung, Pei-Yuan Chen, Xiang-Feng Hong, Jia-Wei Liu, Yi-Heng Li 

1 Graduate Institute of Hydrologic and Oceanic Sciences, National Central University, Taoyuan, Taiwan,

2 Department of Earth Sciences, National Central University, Taoyuan, Taiwan,

3 ABEL TECHNOLOGY CONSULTANTS CO., LTD., Taoyuan, Taiwan,

4 Green Energy and Environment Research Laboratories, Industrial Technology Research Institute, Hsinchu 31057, Taiwan

 

     Even under the same rainfall conditions, soil moisture in different lands may vary greatly, and soil moisture affects vegetation growth and requires different amounts of irrigation water. Therefore, predicting future soil moisture changes is very helpful for water resource management and agriculture. This study utilizes water balance models and machine learning methods to predict soil moisture content. The first is based on green roofs developed in the past, trying to simulate soil moisture in grasslands, dividing historical data into independent time periods, and simulating the process of increasing soil moisture during rainfall. The second method uses the random forest method and inputs the rainfall to estimate the soil moisture content in the next week or two weeks.

​

​