Water Watch Hackathon

Forecasting the future water availability

The city of Bangalore, India, faces challenges to provide enough water for the current and future water demand for drinking and irrigation purposes. The wet and dry periods are becoming more extreme, which together with the rise of population results in the need for a better insight of the future water availability in the four main reservoirs around Bangalore.

A Quantile Regression approach (ML) provides us with the uncertainty bands of the future water availability forecast. The machine learning technique used is trained for monthly, weekly and daily forecasts for the reservoirs inflow, based on incoming radiation and precipitation forecasts.