Seasonal prediction of wind and solar PV generation in India during the western summer monsoon

India's ambitious net-zero climate goals include plans for significant increases in wind and solar photovoltaic (PV) energy generation by 2030. Despite the importance of these two sources in India's future energy mix, few studies have estimated the magnitude of interannual generation variability that results from such expansion. Furthermore, there is little understanding of the meteorological drivers of wind and solar PV generation on seasonal timescales, nor any documented attempts to use dynamical seasonal forecasts to predict seasonal generation anomalies.

This thesis explores interannual variability in India's wind and solar PV generation, focusing on the summer monsoon season (June, July, August, and September). An atmospheric reanalysis is used to create synthetic wind and solar PV generation time series for India to study variability. And data outputs from a seasonal forecasting system are used to test seasonal generation predictions for wind and solar PV.

The generation timeseries wind and solar PV were found to be an accurate means of representing observed generation variability on a range of timescales. Relationships between anomalies in the multi-decadal generation syntheses and meteorological drivers were investigated. The western North Pacific Monsoon and Indian summer monsoon circulations were found to be responsible for the majority of observed interannual variability in wind and solar PV generation in summer, respectively. Skilful forecasts of summer season wind and solar PV generation at a one-month lead time are possible using seasonal climate forecasts initialised in May.

The generation syntheses were found to be sensitive to the technology parameterisations of wind turbines and solar arrays. Additional generation syntheses that used state-of-the-art turbine and solar array parameterisations appreciably increased the capacity factors of both technologies when rolled out nationwide. This suggests that less installed capacity is required to achieve specific generation outcomes than previously estimated.