Abstract:
An innovative approach for drought identification is developed using Multi-Criteria
Decision-Making (MCDM) and Artificial Neural Network (ANN) model from surveyed
drought parameters data around the Dhalai river watershed in Tripura hinterlands, India.
Total eight drought parameters i.e. precipitation, soil moisture, evapotranspiration, vegetation
canopy, cropping pattern, temperature, cultivated land, and groundwater level were obtained
from experts, literature and cultivators survey. Then, the Analytic Hierarchy Process (AHP)
and Analytic Network Process (ANP) were used for weighting of parameters and Drought
Index Identification (DII). Field data of weighted parameters in the meso scale Dhalai river
watershed were collected and used to train the ANN model. The trained ANN model has been
tested in the same watershed for its calibration. Results indicate that the Limited Memory -
Quasi Newton algorithm was better than the commonly used training method. Based on
obtained results from ANN model drought index 0.30 to 0.75 were generated for present
study area. Overall analysis revealed that, with appropriate training, the ANN model could be
used in the areas where the model is calibrated, or other areas where range of input
parameters is similar to the calibrated region.