An Assessment of the ANFIS and GP techniques in Predicting Hydrological Drought from Meteorological Drought for the Sufi-Chai catchment

Hydrological drought (HD) is the most devastating natural phenomena that are falls the inhabitants of dry lands as its dive consequences may not be compensated. Therefore، its prediction has been puzzling researchers from time immemorial. As this phenomenon is directly related to the meteorological drought (MD) we decided to establish a relationship between the two for the Sufi-Chai Catchment، the Province of East Azarbaijan، benefitting from 40 years (1950-1989) of monthly precipitation and stream flow data. Using the normal monthly regime method، the drought periods data relating to each series of precipitation and stream flow were extracted and their relationship were established using the GP and ANFIS artificial intelligence techniques in order to predict the occurrence of drought. The accuracy and precision of these 2 techniques were evaluated using the correlation coefficient (r2)، root mean square error (RMSE) and mean absolute deviation (MAD). The very high r (0. 99) indicated that both techniques correctly predicted the temporal occurrence of HD from MD. The results indicated that ANFIS with the RMSE=4. 98 and MAD of 3. 83 was superior to GP with the RMSE=6. 437 and MAD of 5. 021.