Abstract The Arabian Sea experiences an intense perennial Oxygen Minimum Zone (OMZ) (150–1,200 m) in its northern and central regions. Earlier measurements during the 1990s (Joint Global Ocean Flux Study, India) revealed methane (CH 4 ) oversaturation in the upper 300–400 m in this region. The basin is reported to have experienced warming and OMZ intensification in recent years, while the particle flux is expected to remain moderately steady. In response to the above phenomena, we aimed to examine the change in CH 4 distribution in comparison to the 1990s by studying the depth profiles of CH 4 along 8–21°N over 68°E meridian from February 2017 to October 2024. The present study showed CH 4 oversaturation in the upper 135–350 m in the OMZ region and 100–150 m in the non‐OMZ region. The upper OMZ CH 4 maxima exhibits moderate spatial variability with higher concentrations toward the north. However, there was no significant change in the CH 4 maxima in the OMZ in comparison to those during the 1990s. The waters in the OMZ core and below were undersaturated with CH 4 as observed in the 1990s. We hypothesize that the CH 4 build‐up in the OMZ waters was possibly related to its in situ production from the anoxic micro‐niches in sinking detrital particulate matter. The CH 4 undersaturation in the core of OMZ was possibly due to reduced particle fluxes, which led to low CH 4 production potential. However, the potential role of oxidants such as nitrite and oxygen on the CH 4 cycling in the Arabian Sea OMZ needs further research.
Abstract Machine Learning Weather Prediction (MLWP) models have recently demonstrated remarkable potential to rival physics‐based Numerical Weather Prediction (NWP) models, offering global weather forecasts at a fraction of the computational cost. However, thorough evaluations are essential before considering MLWP models as replacements for NWP models. This study presents a comprehensive evaluation of four leading MLWP models—GraphCast, PanguWeather, Aurora, and FourCastNet—against observations and three state‐of‐the‐art NWP models in predicting tropical cyclones (TCs) across all tropical ocean basins. All MLWP models exhibited strong skill in forecasting TC tracks, achieving an average track error of less than 200 km at a 96‐hr forecast lead time. However, they consistently underestimated maximum sustained wind speeds compared to NWP models and observations. The low bias in TC intensity forecasts by MLWP models is linked to similar bias in their training data, along with the double penalization effect. MLWP models realistically captured the absolute vorticity patterns and their advection, demonstrating their ability to represent the dynamics underlying TC translation. They also captured the low‐level convergence and vertical warm core structure of TCs, although the magnitudes were weaker than observed, highlighting the linkage between dynamical and thermodynamical processes. The consistency in magnitude between various physical fields in the MLWP models suggests that they intuitively learn the interrelationships among different physical fields during the evolution of weather systems, demonstrating their ability to capture complex physical interactions. Among the MLWP models, Aurora showed superior performance, surpassing GraphCast, PanguWeather, and FourCastNet.