search for


Comparison of quantitative precipitation estimate using geostatistical models
Journal of the Korean Data & Information Science Society 2019;30:77-89
Published online January 31, 2019;
© 2019 Korean Data and Information Science Society.

Jieun Im1 · Sanghoo Yoon2

1Department of Statistics, Daegu University
2Division of Mathematics and Big Data Science, Daegu University
Correspondence to: Assistant professor, Division of Mathematics and Big Data Science, Daegu University, Gyeongbuk 68453, Korea. E-mail:
Received December 3, 2018; Revised January 2, 2019; Accepted January 2, 2019.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License ( which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
Precipitation is necessary data for hydrological studies to reduce the damage caused by meteorological disasters such as heavy rainfall. There are being conducted studies of combining gauge data and radar data to improve quantitative precipitation estimate. The advantage of radar is monitoring the rainfall phenomenon over a large area and can estimate the rainfall at the non-measurement sites. This study considered geostatistical models for quantitative precipitation estimate. The models used are inverse distance weight, generalized additive model, kriging, and spatial random forest. To evaluate the prediction performance, 10-folds cross-validation was repeated 50 times and root mean square error, mean absolute error, relative bias is calculated. As a result, inverse distance weight performs best when rainfall was low and spatial range was local. In the opposite situation, ordinary kriging performs best. A spatial random forest was regarded as robust predictors regardless of rainfall intensity. Spatial random forest performs well when time resolution is high. however ordinary kriging performs best for low time resolution.
Keywords : Generalized addictive model, kriging, spatial random forest.