Title: Application of global precipitation dataset for drought monitoring and forecasting over the Lake Urmia basin with the GA-SVR model

Authors: Edris AhmadEbrahimpour; Babak Aminnejad; Keivan Khalili

Addresses: Department of Civil Engineering, Roudehen Branch, Islamic Azad University, Roudehen, Iran ' Department of Civil Engineering, Roudehen Branch, Islamic Azad University, Roudehen, Iran ' Department of Water Engineering, Urmia University, Urmia, Iran

Abstract: In the present study, the accuracy of the climate research unit (CRU) precipitation data was assessed as an alternative source instead of in situ data for monitoring the drought in the Lake Urmia Basin area during the period from 1984 to 2013. Later, a genetic algorithm-support vector regression (GA-SVR) model was utilised in order to forecast drought conditions up to four months ahead. The results demonstrated that the CRU data had acceptable accuracy in drought monitoring so that in at least 75% of the cases, there was no difference between the monitored drought classed through observed data and CRU data. In the forecasting section, the results showed two general patterns. The first pattern indicated a descending trend of forecast accuracy with an increase in the lead-times ahead of forecasts; the second pattern revealed the ascending trend of forecast accuracy, with an increase in the SPI scale.

Keywords: climatic research unit; CRU; genetic algorithm; support vector regression; gridded precipitation.

DOI: 10.1504/IJW.2018.093674

International Journal of Water, 2018 Vol.12 No.3, pp.262 - 277

Received: 02 Dec 2017
Accepted: 09 Mar 2018
Published online: 31 Jul 2018 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article