Improving the forecast quality of near-term climate projections in the Mediterranean region

The Mediterranean region is often considered a climate change hotspot. State of the art climate projections display a large uncertainty due to factors such as the modeling approach or variability phasing, in particular for projections of the next few decades. With the great amount of climate information from the last Coupled Model Intercomparison Project (CMIP6), it is desirable to constrain the multi-model multi-member ensemble for the near-term future to obtain more robust and better performing projections. We explore different subsetting methods that select those members that better capture the variability at the starting date of the projection’s study period. To find the best information we explore variations of different parameters in the selection method based on relevant climate drivers in the Mediterranean region. Such parameters are the reference against which the best members are selected, and comprise: the time period used, the constraining regions considered and the variable or metric that drive the constraint. We find that these subsetting methods can improve the accuracy of near-term climate projections for the next 20 years compared to the unconstrained projections. This evaluation of the results allows us to make informed and robust decisions about the near-term future projections based on quality estimates borrowed from climate prediction practices.