Multischeme ensemble forecasting of surface temperature using neural network over Turkey

Cakir S., Kadioglu M., CUBUKCU N.

THEORETICAL AND APPLIED CLIMATOLOGY, cilt.111, ss.703-711, 2013 (SCI-Expanded) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 111
  • Basım Tarihi: 2013
  • Doi Numarası: 10.1007/s00704-012-0703-1
  • Dergi Adı: THEORETICAL AND APPLIED CLIMATOLOGY
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Sayfa Sayıları: ss.703-711
  • İstanbul Teknik Üniversitesi Adresli: Evet

Özet

The ensemble method has long been used to reduce the errors that are caused by initial conditions and/or parameterizations of models in forecasting problems. In this study, neural network (NN) simulations are applied to ensemble weather forecasting. Temperature forecasts averaged over 2 weeks from four different forecasts are used to develop the NN model. Additionally, an ensemble mean of bias-corrected data is used as the control experiment. Overall, ensemble forecasts weighted by NN with feed forward backpropagation algorithm gave better root mean square error, mean absolute error, and same sign percent skills compared to those of the control experiment in most stations and produced more accurate weather forecasts.