My research uses machine learning to produce skillful, trustworthy climate predictions. With intentional experimental design and interpretable AI techniques, machine learning can produce new insights into the behavior of the climate system. Check out some of my interests below.
Detecting the Climate Change Signal
Separating internal variability and the externally forced response is necessary for understanding the behavior climate change. Patterns of internal variability and forced change can take on similar structures (e.g., El Niño-Southern Oscillation and El Niño-like warming), making this an ongoing challenge in climate science.
Identifying Predictable Climate States
While chaos is inherent to the atmosphere and ocean, some climate states offer skill in predicting future climates states. I specialize in developing interpretable machine learning architectures that reveal these precursor patterns.
Forecasting the Near-Term Climate
Hydrology, energy, and fisheries are among the many sectors that benefit from seasonal-to-decadal forecasts. Data-driven methods can utilize the immense amount of simulated climate data to produce refined forecasts on these time scales.
Jamin Rader 