Quantifying uncertainty in twenty-first century climate change over India

Abstract

Uncertainties in future changes of temperature and precipitation over the homogenous monsoon regions of India are investigated using the CMIP5 and CESM-LE datasets. The uncertainty is partitioned into epistemic (model) and aleatoric (internal variability) components for each season using the RCP8.5 scenario. The uncertainty in temperature change is dominated by epistemic uncertainty that increases over time. The uncertainty in precipitation change shows a more complex picture. Aleatoric uncertainty can remain quite large and comparable to epistemic uncertainty till the latter part of the twenty-first Century especially during the JJA and SON seasons. Much of the rainfall uncertainty is in the more arid Northwest region with the West Central region (part of the core monsoon area) exhibiting lower uncertainties. Considerable increase in rainfall is seen during the SON season indicating an extended monsoon season. During the DJF season aleatoric uncertainty is much larger than epistemic uncertainty over much of the century and shows considerable decadal scale variability. Using the 40-member CESM-LE ensemble to analyze the influence of ensemble size on aleatoric uncertainty we find that low ensemble sizes can lead to an underestimate of the uncertainty.