Unravelling large-scale onset and progression of the Indian monsoon from the evolution of clusters of local onsets using network science

Developing an accurate definition of the onset and progression of the Indian monsoon is an outstanding research area in climate science. Determining monsoon onset dates is critical for agricultural planning and ensuring food security for billions of people in India and the world. The onset of the Indian monsoon is associated with the northward shift of the planetary-scale intertropical convergence zone (ITCZ) from the equator. ITCZ is a zone of intense convective activity, cloudiness and high precipitation girdling the Earth. As a result, the onset and progress of the Indian monsoon are interconnected on a planetary scale.

The monsoon onset definition can be classified into local and large-scale definitions. Local-scale definitions utilize daily precipitation over a small region to determine the onset. However, they are prone to bogus onsets because of pre-monsoon rains and transient weather systems. Large-scale definitions are based on precipitation and wind/cloudiness over a bigger area. However, such averaging does not guarantee separation of the ‘large-scale’ component of ITCZ precipitation from the ‘small-scale’ local contributions and is still prone to bogus onsets. Large-scale onset definitions are largely confined to defining the monsoon onset over Kerala (MoK) while representing the progression of monsoon is based entirely on local onsets. We overcome this limitation by developing a large-scale definition from small-scale local onsets interconnected on a planetary scale.

We utilize networks and their phase transitions to develop a large-scale definition. We construct time-varying spatial proximity networks based on daily precipitation, where nodes are the geographical locations in a domain encompassing India. Links are established only between nodes that are in geographical proximity and if they have undergone local onsets. Next, we estimate connected components in the network that represent clusters of local onsets. The spatiotemporal evolution, involving the growth and coalescence of clusters disentangles the true large-scale monsoon onset and progression.

We discover two abrupt phase transitions in the size of the largest cluster of the local onsets. These phase transitions are associated with the formation of large clusters representing the local onsets interconnected at a planetary scale. Thus, we unravel the setting up of the ITCZ and other synoptic-scale convective systems that facilitate consistent monsoon activity over India. We define large-scale onsets when a location becomes part of the largest cluster following the first transition.

Using lead-lag composites of precipitation for the past 84 years centred on large-scale onsets, we find that our definition captures important characteristics of the Indian monsoon usually missed by conventional large-scale definitions. During our onsets, the rainfall is strong at a large scale along the western ghats and northeast India (NEI). Further, they are followed by a rapidly northward propagating rainfall pulse, also known as the monsoon intraseasonal oscillation. Our method captures that the onset over NEI occurs before MoK, which is consistent with several recent studies but missed by conventional definitions. These new findings necessitate a reexamination of the interannual variability in the Indian monsoon, which will be discussed in the talk.