A hybrid model for statistical downscaling of daily rainfall

The robustness of random forest (RF) in classification and superiority of support vector machine (SVM) to fit highly non-linear data were used to develop a hybrid model for statistical downscaling of daily rainfall. The RF was used to predict whether rain will occur in a day or not and SVM was used...

Full description

Saved in:
Bibliographic Details
Main Authors: Pour, S. H., Shahid, S., Chung, E. S.
Format: Conference or Workshop Item
Published: Elsevier Ltd 2016
Subjects:
Online Access:http://eprints.utm.my/id/eprint/73639/
https://www.scopus.com/inward/record.uri?eid=2-s2.0-84997770183&doi=10.1016%2fj.proeng.2016.07.514&partnerID=40&md5=c6b9a0478471b6892679cb1ea6312149
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The robustness of random forest (RF) in classification and superiority of support vector machine (SVM) to fit highly non-linear data were used to develop a hybrid model for statistical downscaling of daily rainfall. The RF was used to predict whether rain will occur in a day or not and SVM was used to predict amount of rainfall in rainfall occurring days. The capability of proposed hybrid model was verified by downscaling daily rainfall at three rain-gauge locations in the east cost of peninsular Malaysia. Obtained results reveal that the hybrid model can downscale rainfall with Nash-Sutcliff efficiency in the range of 0.90-0.93, which is much higher compared to RF and SVM downscaling models. The hybrid model was also found to replicate the variability, number of consecutive wet days, 95-percentile rainfall amount in each months as well as distribution of observed rainfall reliably.