A novel constraining method for a better description of the Indian monsoon precipitation change

Beyond future emission pathways, projections of precipitation in a changing climate are still showing a large spread among CMIP's Global Circulation Models (GCMs), especially at the regional scale. This is mainly arising from the so-called model uncertainty, i.e. from our limited knowledge in but also from the plural representation of climate system's complex mechanisms. Those uncertainties represent a point of great concern for the design of responsible regional adaptation policies, and it is urgent to reduce these uncertainties to better assess future change in regional precipitation. While ongoing and future improvement of GCMs will surely allow for precision of climate change trajectory, here we suggest to make the best use of already existing information for uncertainty reduction now.

We will focus on the example of the Indian summer monsoon, being a regional phenomenon of importance for the livelihood of billions of people; yet its evolution under climate change is largely uncertain. Using the two latest generations of GCMs (CMIP5 and CMIP6), we suggest an original method for constraining models' projections of Indian summer precipitation change based on observations and using an inter-model Maximum Covariance Analysis (MCA) technique. Our method is compared to a straightforward emergent constraint approach and shows  promising and robust results, both in terms of reduction in uncertainty and in the explanation of underlying physical mechanisms. Additionally, a robustness assessment is done through a perfect model validation i.e. by checking the ability of our method to reliably predict a left one out model. We believe robustness-checks are a needed procedure for an honest and trustworthy reduction in uncertainty of future change.