By Jon Shonk
Over one billion people are reliant on the rainfall of the Indian Summer Monsoon. During the wet season, which usually spans June to September, some parts of India receive over 90% of their total annual rainfall. Deficits or excesses of rainfall can have devastating effects, such as drought, inundation, crop failure and health issues. Bouts of extreme weather, such as short periods of very intense rain, can also have detrimental effects via flash flooding and the triggering of landslides.
It is therefore important to get an idea of how monsoon rainfall might change in a warmer future climate. Climate prediction uses numerical models to advance an initial global “snapshot” of the atmosphere and ocean forward in time using a supercomputer, and then examines the statistics of the weather over some period in the future. Nowadays, many institutions around the world run their own climate models. While these are all based around the same physical principles, the formulation and structure of the models can be very different. This means that the behaviour of models, even if initialised from the same global snapshot, can be quite different after 100 years of simulation.
The five maps in Figure 1 show the projected change in rainfall (averaged from June to August) over India in a future world that is 1.5 °C warmer than pre-industrial conditions (about 0.8 °C warmer than today), for five different climate models. There is a clear disagreement in the pattern of change, with no obvious consensus on which parts of India are likely to become wetter or drier.
Figure 1. Projected changes in rainfall as a result of a 1.5 °C warming, according to five climate models. Rainfall is averaged over June, July and August. Data from the HAPPI project (see Mitchell et al, 2017 for details).
So can we infer anything about the future Indian Summer Monsoon from these models? An advantage of using multiple models is that we can build an “ensemble” of predictions – that is, a number of plausible future climate projections. But the challenge is then how to statistically combine the projected changes to produce a single, clear, robust message.
The simplest option is to take an average across the projections from the five models (Figure 2a). The result is a weak pattern of slightly wetter conditions over eastern India and Bangladesh. However, the averaging process leads to areas where an increase of rainfall in one model cancels out a decrease in another, and understanding the reasons why models project such differences could provide extra clues as to how the monsoon might change.
Figure 2. Projected changes in rainfall, shown as the average across the same five models used in Figure 1. The changes for a warming of (a) 1.5 °C and (b) 2.0 °C are shown. Data from the HAPPI project.
By examining the behaviour of the models individually we can build an idea of the mechanisms by which the rainfall distribution changes in a warmer climate. This has been the focus of my recent work. I have also been looking at the differences in rainfall change between a world that is 1.5 °C warmer and one that is 2 °C warmer (Figure 2b). A paper on this should be ready soon…