Publications

Khosravi, I; Jouybari-Moghaddam, Y; Sarajian, MR (2017). The comparison of NN, SVR, LSSVR and ANFIS at modeling meteorological and remotely sensed drought indices over the eastern district of Isfahan, Iran. NATURAL HAZARDS, 87(3), 1507-1522.

Abstract
This paper aims to employ and compare four methods of neural network (NN), support vector regression (SVR), least squares support vector regression (LSSVR) and adaptive neuro-fuzzy inference system (ANFIS) for modeling the time series behavior of the meteorological and the remotely sensed (RS) drought indices of the eastern district of Isfahan during 2000-2014. The data used in the paper are the normalized difference vegetation index (NDVI) and the land surface temperature time series of MODIS satellite and the rainfall time series of TRMM satellite. Then, three RS drought indices namely vegetation condition index, NDVI deviation index and temperature vegetation index and three meteorological drought indices namely 3-month SPI, 6-month SPI and 12-month SPI are generated by the data. Afterward, based on the correlation coefficient between the RS and the meteorological drought indices, three indices are chosen as candidate indices for monitoring the drought severity of the study area. After modeling the time series behavior of these indices by the aforementioned methods, the results indicate that the SVR has the highest and the NN has the lowest efficiency among all the methods. In addition, the performance speed of the LSSVR and then the ANFIS is the highest. At the end of the paper, a fuzzy inference system (FIS) is presented based on the candidate indices to monitor the drought severity at spring and summer of 2000-2014. According to the results of the designed FIS, the spring status is normal in all years except 2000 and 2011 (moderate drought) and the summer status is severe drought in all years except 2000, 2010, 2011 and 2014 (moderate drought).

DOI:
10.1007/s11069-017-2827-1

ISSN:
0921-030X