Research report: Radiative forcing variability and its moderation during a cyclonic event: a case study

Severe Cyclone Mocha sharply altered the aerosol composition over the Bay of Bengal in May 2023, raising concentrations of wind-generated sea salt and dust and producing a regional radiation response dominated by shortwave cooling rather than longwave heating. A case study combining atmospheric reanalysis, satellite observations and radiative-transfer modelling found that aerosol optical depth exceeded 1 over parts of the storm’s mature core and landfall region. Model and satellite analyses indicated cooling at the top of the atmosphere and at the surface, accompanied by warming within the atmospheric column.

The work examined one cyclone, one event period and one regional domain, rather than establishing a general response for all tropical cyclones. The analysis covered 4–19 May 2023, including background conditions before Mocha began developing on 9 May, its mature phase over the Bay of Bengal, landfall on 15 May and subsequent dissipation. The study domain extended from 6°N to 22°N and from 84°E to 94°E, covering the Bay of Bengal and adjoining areas of Bangladesh, Myanmar and eastern India, including West Bengal.

The central questions were how sea salt, mineral dust and black carbon changed during the storm, how aerosol optical depth reflected those changes, and how the altered aerosol population affected shortwave and longwave direct radiative forcing at the top of the atmosphere, the surface and within the atmosphere. Daily aerosol and meteorological fields came from the Modern-Era Retrospective analysis for Research and Applications, Version 2, or MERRA-2, a global reanalysis produced from numerical weather prediction models and observations. The dataset supplied mass concentrations of sea salt, dust and black carbon, aerosol optical depth at 550 nanometres, and wind data at 10 metres and 850 millibars.

MERRA-2 aerosol optical depth was compared with retrievals from the Moderate Resolution Imaging Spectroradiometer aboard the Terra and Aqua satellites. Those ocean retrievals used a dark-target algorithm at a spatial resolution of 1° by 1°. The comparison also incorporated MODIS cloud properties, including cloud fraction, cloud-top pressure, cloud-top temperature, liquid-cloud optical thickness, cloud effective radius and liquid-cloud water path.

Radiative effects were estimated with the Santa Barbara DISORT Atmospheric Radiative Transfer model, or SBDART, a one-dimensional radiative-transfer model that accounts for atmospheric absorption, scattering and surface reflection through the DISORT algorithm. The Optical Properties of Aerosols and Clouds model generated optical parameters for a mixture of water-soluble particles, accumulation- and coarse-mode sea salt, and accumulation- and coarse-mode mineral aerosols. To provide these inputs, MERRA-2 aerosol mass concentrations were converted to number concentrations.

Shortwave calculations covered wavelengths from 0.3 to 3 micrometres, while longwave calculations covered 4 to 100 micrometres. Researchers checked the SBDART results against daily radiation fluxes from the Clouds and the Earth’s Radiant Energy System, or CERES. To separate changes associated with aerosols from those associated with clouds, the researchers ran a comparison using 13 May aerosol conditions with the cloud conditions specified for 4 May, then evaluated that configuration against the 4 May baseline.

This was not simply a direct comparison of all atmospheric conditions on the two dates: it held the earlier cloud state fixed while replacing the aerosol state with the mature-storm distribution. That design was intended to estimate aerosol-only forcing. The combined effects of cloud and aerosol changes were assessed separately, allowing the study to compare their relative magnitudes.

This distinction was important because cloud cover and cyclone evolution changed substantially during the event. Sea salt provided the clearest evidence of cyclone-associated aerosol generation. Its concentration was approximately 100 micrograms per cubic metre on 4 May, before cyclonic activity developed, but rose to about 2,400 micrograms per cubic metre near the cyclone core and surrounding waters on 11–13 May.

High concentrations first appeared around the core on 5–9 May and subsequently spread across much of the Bay of Bengal as the cyclone intensified from 10 to 13 May. The authors linked this increase to the exceptional wind stress generated over the ocean, which promotes sea-spray production through bubble bursting. Immediately after landfall, the concentration fell to roughly 250 micrograms per cubic metre and later returned to about 100 micrograms per cubic metre during the dissipation phase.

The spatial and temporal distribution therefore followed the cyclone’s path rather than remaining at a fixed background level. Dust displayed a different, land-centred pattern. Before the cyclone began, concentrations over the ocean and surrounding land were only about 1–5 micrograms per cubic metre, and little changed over the water when cyclonic activity began on 7 May.

Higher concentrations appeared over bordering land as the system approached, reaching roughly 60 micrograms per cubic metre. Around and immediately after landfall, dust reached about 100 micrograms per cubic metre over land and nearby coastal waters, and the elevated concentrations persisted for several days. The study associated this increase with strong winds capable of raising soil erosion and with post-landfall soil disturbance.

Black carbon behaved differently: its concentration was generally near or below 1 microgram per cubic metre, with only a slight increase over neighbouring land as winds transported particles and no discernible major change over the ocean. The strongest shifts thus involved the naturally generated sea salt and dust components rather than the anthropogenic black-carbon aerosol. The redistribution of these particles was visible in aerosol optical depth, an indicator of atmospheric aerosol loading that reflects particle abundance and composition.

Sea salt dominated marine loading during the cyclone, while dust became more important over land and coastal waters during the dissipation stage. Values exceeded 1 over the mature cyclone core on 13 May and over a wider area during landfall on 15 May. MODIS generally confirmed the spatial pattern obtained from MERRA-2, although MODIS values were slightly higher on 8–10 May and again on 16–19 May, when coastal dust concentrations were elevated.

The comparison indicated that MERRA-2 underestimated optical depth under dusty conditions. Cloud cover caused some pixels to be unavailable for MODIS retrieval, although the extent of these gaps should not be confused with a complete absence of observations. Even the cyclone eye could lack retrievals because the algorithm applied cloud filtering and pixel-selection criteria.

Shortwave radiation accounted for the larger modelled aerosol-induced change. In the SBDART area-average estimate for 4 May, shortwave forcing was about −16.42 watts per square metre at the top of the atmosphere and −21.66 watts per square metre at the surface. By 13 May, the corresponding area averages became more negative, reaching −17.31 and −22.81 watts per square metre, indicating greater shortwave cooling.

Within the cyclone core, individual SBDART estimates fell to roughly −25 watts per square metre at the top of the atmosphere and below −30 watts per square metre at the surface. CERES resolved a larger region of reduced shortwave forcing than the one-dimensional model but found a similar central pattern. Its cloud-and-aerosol difference reached −7.73 watts per square metre at the top of the atmosphere and −8.69 watts per square metre at the surface.

With cloud effects held fixed, the SBDART aerosol-only shortwave changes were −0.168 watts per square metre at the top of the atmosphere and −0.161 watts per square metre at the surface; the corresponding CERES estimates were larger at −1.00 and −1.64 watts per square metre. Longwave forcing moved in the opposite direction but remained weaker. SBDART area-average estimates for 4 May were 4.13 watts per square metre at the top of the atmosphere and 2.44 watts per square metre at the surface.

On 13 May, these increased to 4.35 and 2.57 watts per square metre. At the cyclone core, the top-of-atmosphere longwave forcing reached about 6 watts per square metre, with values of 4.3–6 watts per square metres across the broader cyclone region. The cloud-and-aerosol difference was +0.22 watts per square metre at the top of the atmosphere and +0.13 watts per square metre at the surface in SBDART, while CERES estimated smaller increases of +0.09 and +0.07 watts per square metre.

The SBDART aerosol-only changes were +0.026 watts per square metre at the top of the atmosphere and +0.017 watts per square metre at the surface; CERES values were +0.0101 and +0.0091 watts per square metre. Both approaches captured the same sign and order of magnitude, although CERES generally estimated less longwave heating than SBDART. The regional balance was dominated by shortwave effects, which the study relates to the increasing prominence of fine-mode particles as the cyclone intensified.

The authors note that larger particles are removed by rain more rapidly than finer particles, while fine-mode particles scatter visible solar radiation more strongly and are comparatively transparent to longwave radiation. The area-average calculations therefore indicated cooling at the top of the atmosphere and surface, with warming retained within the atmospheric column. These findings do not show that aerosols determined Mocha’s intensity or that radiation was the cyclone’s dominant process.

In the model comparisons, changes associated with clouds were roughly an order of magnitude greater than aerosol-only changes. The analysis was also limited by its one-storm scope, the simplified vertical structure represented by a one-dimensional radiative-transfer model, missing satellite retrievals in some cloud-affected pixels, and differences between CERES and SBDART in magnitude and spatial extent. Within those constraints, the case study documents a rapid shift toward sea salt and dust loading, followed by stronger shortwave cooling, weaker longwave heating and a net cooling tendency at the top of the atmosphere and surface during the mature phase and landfall.

Subject of Research: Aerosol redistribution and radiative forcing during severe Cyclone Mocha in the Bay of Bengal in May 2023

Article Title: Radiative forcing variability and its moderation during a cyclonic event: a case study

Article References: Srivastava, N. (2026). Radiative forcing variability and its moderation during a cyclonic event: a case study. Frontiers in Environmental Science, 14, 1924384. https://doi.org/10.3389/fenvs.2026.1924384

Image Credits: AI Generated

DOI: 10.3389/fenvs.2026.1924384

Keywords: Cyclone Mocha, Bay of Bengal, aerosol optical depth, radiative forcing, sea-salt aerosol, dust aerosol, shortwave radiation, longwave radiation, MERRA-2, satellite observations, Radiative, forcing