Robustness of precipitation Emergent Constraints in CMIP6

Climate models are fundamental tools to understand the complexity of the climate system, to study  the processes at work and to provide credible future climate projections. Unfortunately, models often disagree significantly in the amplitude of different climate change signals and in their representation of the role of important feedbacks. In the past years the “Emergent Constraints” methodology has been developed for reducing uncertainties in climate-change projections. An Emergent Constraint (EC) is a statistical relationship between the inter-model spread of a measurable aspect of the present-day climate (predictor) and the inter-model spread of a variable projection (predictand), under a climate change scenario. If a significant correlation is found, observations of the predictor can be used to constrain model projections of the predictand and the uncertainties in climate model outputs can be narrowed. 

In the last two decades, ECs have been identified in different branches of climate science although just a limited number of these ECs is related to the hydrological cycle. Recently, a relevant number of EC in the literature was discovered to lack a satisfying physical explanation and many, developed and tested with CMIP5 ensemble, seem to be not-significant in the new CMIP6 ensemble. However, the analysis regarding ECs related to the hydrological cycle is still incomplete. The aim of this work is to test three ECs related to mean-precipitation and extreme precipitation events, originally identified in CMIP3 or CMIP5 data, and to evaluate if their statistical significance survives also in the CMIP6 ensemble: (a) global hydrological sensitivity used to constrain future changes in local extreme precipitation: we find this relationship not to be robust in CMIP6 models; (b) future changes in the Indian summer monsoon precipitation, constrained by Western Pacific mean precipitation: this relationship is not robust with the new ensemble; (c) changes in future extreme tropical precipitation, constrained by the same variable calculated in the past: we find this EC to be robust both in CMIP5 and CMIP6.