DATE2022-05-16 16:29:38
TITLEConstraining CMIP6 projections in the Mediterranean region: on the road towards user-oriented climate information
AUTHORSJosep Cos (0) ,Francisco Doblas-reyes (0) ,Raül Marcos (0) ,Nube González-reviriego (0)
ABSTRACTThe Mediterranean region is considered to be a climate change hotspot due to an observed and projected increase in warming trends and precipitation declines. In a rapidly changing climate, reliable climate information for the next 20 years is essential for countries located around the Mediterranean basin. However, with the amount and variety of climate prediction and projection data available, it is often hard to make decisions about the quality of the climate information generated. To overcome this issue, different constraining methods exist that focus on reducing the uncertainty of multi-model ensembles such as the ones produced by the Coupled Model intercomparison Projects (CMIP). Specifically, the latest CMIP exercise (CMIP6) is known to have produced much warmer projections, and it is argued that the climate sensitivity of some models is beyond the likely range. Therefore, we believe some treatment to the raw projections from CMIP6 is needed to provide more robust estimates of the future climate. To this aim, three different constraining methods have been used to compare their performance. Two methods are based on sub-selecting the ensemble members that are in better agreement with the observed variability, one of them using decadal predictions and the other one observations for the member selection. The third method uses individual weights for each member based on their performance against the observed climate and their independence from the rest of the ensemble members. The three methods are implemented to estimate the state of the climate up to 20 years in advance. Therefore, they can be used to compute climate projections in "hindcast mode" over periods for which there are observations available and, hence, an estimation of the quality of the information can be provided for different essential variables. Focusing on 20-year projections helps to bridge the information gap given by decadal predictions and raw projections. Consequently, by constraining the future projections using different approaches, we aim to gain confidence in the estimates of the near-term changes expected in the Mediterranean region. This is important to be able to provide more reliable user-oriented climate information, which reinforces its potential applicability for impact studies as, for example, the provision of tailored bioclimatic indicators such as those relevant for the wine sector.