Climate Change Impact on Virtual Water Availability: A Categorized Polynomial Neural Network Approach

The uncontrolled extraction followed by technological advancements and increasing population has induced an increase in Green House Gas concentration in the atmosphere and the outcome was global warming followed by the change in climatic pattern in different places of the world has reduced per capita availability of water. Due to climate change, water accessibility for manufacturing also gets affected. The volume of water which is also avowed as virtual water is significantly impacting the industrial output. Few studies tried to estimate the consequences of variation in climatic conditions change on virtual water utilization but at present, no study has considered the same impact on the availability of virtual water. That is why climate change impact was estimated on the availability of virtual water per consumer of the industrial output of the region in the current study. The estimation was depicted in cardinal rather than ordinal output. Also, the Polynomial Neural Network architecture-based Group Method of Data Handling was utilized as the optimal classifier of climate change impact on virtual water availability per consumer. Four metro cities of the Indian Sub-continent were considered as the study area. According to the results New Delhi is worst and Kolkata least affected for the Intergovernmental Panel on Change of Climate proposed B2 scenario. In the case of the A2 scenario, here also New Delhi was found to be the worst affected and Mumbai least affected. The classifier was found to be more efficient than Stepwise Forward Regression, Decision Forest, and Logistic Regression techniques utilized in the present investigation for the same purpose. But due to the equal sign, that was considered for all the selected input variables the Percentage Correct Classification Rate was found to not above 85% which can be improved by incorporating the importance of the inputs and their effect on the output in the model algorithm.