Prediction of rainfall from WSR-88D radar using kernel-based methods

Theodore B. Trafalis*, Budi Santosa, Michael B. Richman

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

7 Citations (Scopus)


The main objective of this paper is to utilize standard Support Vector Regression, Least Squares Support Vector Regression, and compare these techniques to traditional regression and a rain rate formula that meteorologists use, to facilitate rainfall estimation and rainfall detection. Ground truth rainfall data are necessary to apply intelligent systems techniques. A unique source of such data is the Oklahoma Mesonet. Recently, with the advent of a national network of advanced radars, massive archived data sets are available for data mining. The reflectivity measurements from the radar are used as inputs for the techniques tested. The results show that SVR and LS-SVR are better in terms of generalization error than traditional regression and rain rate formula used in meteorology for both rainfall estimation and rainfall detection. Moreover, LS-SVR shows a better performance than SVR for rainfall estimation and vice versa for rainfall detection.

Original languageEnglish
Pages (from-to)429-438
Number of pages10
JournalInternational Journal of Smart Engineering System Design
Issue number4
Publication statusPublished - Oct 2003
Externally publishedYes


  • Generalization error
  • Kernel function
  • Least squares support vector regression
  • Rainfall estimation
  • Regression analysis
  • Support vector regression


Dive into the research topics of 'Prediction of rainfall from WSR-88D radar using kernel-based methods'. Together they form a unique fingerprint.

Cite this