Using climate model analogs form large ensembles to attribute drought events

Droughts are prolonged periods of water shortages driven by meteorological conditions or human behaviour, affecting social, economic, financial, and eco-systems. These hydroclimatic phenomena can lead to crop failures, diminished water supplies, and negatively impact the environment, posing challenges to food security and economic stability. Meteorological drivers of drought (e.g. precipitation, temperature, radiation patterns) fluctuate due to natural climate variability, but can also be impacted by long-term climatic changes. 

We exploit the concept of climate model analogs to single out specific observed drought events based on the standardized precipitation index, in order to attribute and explain the dynamic drivers that lead to these events months before they occur. In particular the sea surface temperature drivers which remotely modulate hydroclimatic variability over continents in seasonal to annual timescales via teleconnections. This is done by subsampling the most similar events in a large multi-model ensemble to the observed target event and extracting their common signals of prior sea surface temperature conditions and their temporal evolution. In a complementary manner, and as a counterfactual experiment, we predict the standardized precipitation index based on analogs of observed sea surface temperature conditions months before the target drought event and evaluate its skill by comparing predicted values against observed SPI data. 

We have extended the use of an established method of selecting climate model analogs to attribute drought events to internal climate variability. The method can still be further extended to quantify the impact of externally forced climate change versus internal climate variability.