Fog trends in India: Relationships to fog type and western disturbances

Abstract

AbstractFog is a major hazard in wintertime over India, particularly in the Indo‐Gangetic Plains, leading to significant impacts for transport and human health. Using 3‐hourly surface observations, from 69 sites across India, all fog and dense fog events between 2000 and 2020 are identified. For each event, the main fog formation mechanism is objectively categorized using a classification algorithm, distinguishing between radiation, advection, evaporation, precipitation or cloud‐base lowering fog types. In contrast to the findings of other international studies, radiation fog dominates as the most common fog type at the vast majority of locations in India, accounting for 68.1% of all fog events and 70.0% of dense fog events. Statistically significant positive trends are seen in the frequency of all fog events at Delhi, Lucknow and Patna, in the Indo‐Gangetic plains, between 1997/1998 and 2018/2019, dominated by comparable statistically significant positive trends in radiation fogs. Western disturbances (WD) are often linked to the formation of fog in India. Using a climatology of WDs, we show that 46.9% of radiation fog onsets in Delhi in December and January, the primary fog months, happened in conjunction with an active WD event. Conversely, only 32.3% of WDs during these same months coincided with the onset of a radiation fog event. WD‐related radiation fog events are shown to cluster into three distinct groups, with WD centres located to the northwest (51.4% of cases), southwest (13.3%) and east (35.2%) of Delhi. Each cluster is shown to have coherent and distinct near‐surface characteristics which are conducive to fog formation. Trends in WD frequency cannot fully account for the observed trends in fog events. We argue that the fog trends are more likely the result of a complex interaction between urban expansion and the associated rapid change in aerosol loading, resulting in impacts on radiation balance, microphysics and heat‐island processes.