Antarctic icebergs, as a moving source of freshwater while drifting in the Southern Ocean, play a critical role in modulating ocean circulation and sea ice dynamics, especially in light of recent increases in Antarctic ice discharge. Many current ocean and climate models, however, misrepresent Antarctic iceberg melt by assuming it is released along the coast, largely due to the lack of comprehensive assessments that quantify the associated biases and capture the full impacts of this assumption on sea ice and ocean dynamics. This study advances previous work by isolating and quantifying the impacts of iceberg meltwater distribution on the Southern Ocean through the implementation of a more realistic representation of iceberg melt pattern into a global ocean–sea ice model. Compared to traditional coastal-release schemes, the more realistic drifting distribution leads to fresher surface waters, enhanced stratification, and weakened vertical mixing, which reduce upward heat flux, warm the subsurface, and promote the formation of sea ice and Antarctic Bottom Water. Importantly, incorporating the more realistic iceberg meltwater distribution reduces model biases in temperature and salinity climatology by ~ 30% in the high-latitude Southern Ocean. These findings provide quantitative evidence that neglecting the spatial spread of iceberg melt can introduce systematic biases in ocean and climate models.

Smallholder farmers in Sub-Saharan Africa are vulnerable to adverse fluctuations in rainfall, such as dry spells during the critical early stages of the rainy season. In this study, we demonstrate that Malawi is prone to shifts from periods with limited dry spell occurrence to more widespread dry spells later in the season. We develop a predictive model for dry spells, aiming to provide farmers with actionable information to support agricultural decision-making and enhance resilience. The model, based on a dynamical subseasonal prediction system and validated using reanalysis and satellite-based data, focuses on Malawi as a case study. This model has significant skill in predicting the occurrence of at least one dry spell within the three weeks following initialisation, consistently outperforming a climatology-based reference model. Furthermore, we show that the model is applicable beyond Malawi, specifically in East Africa during both the March–May “long rains” and the October–November “short rains”, highlighting its broader relevance for regions where dry spells pose an agricultural risk. The results demonstrate that subseasonal forecasts have the potential to bridge the gap between long-range seasonal outlooks and short-term weather forecasts. Unlike seasonal forecasts, which lack skill at long lead times, subseasonal predictions offer both a longer planning horizon than weather forecasts and greater skill in capturing dry spell risks at actionable lead times. By integrating subseasonal forecasts into national climate services, policymakers and agricultural extension services could provide more timely and targeted advice, potentially helping to mitigate the most severe impacts of dry spells on food production and rural livelihoods.

Abstract Variation of sea surface temperature (SST) results from a host of mechanisms involving both the ocean and atmosphere. The atmospheric forcing on SST is usually hypothesized as white noise to reflect the transient and chaotic nature of the atmosphere in contrast to the large heat inertial of the ocean mixed layer. This study presents evidence that the non-white-noise memory of atmospheric forcing has a substantial contribution to SST variability. According to the computed decorrelation time scale, atmospheric forcing can persist for two to five months in the extratropical North Pacific and North Atlantic. Without considering this atmospheric memory, white noise atmospheric forcing underestimates the observed extratropical SST variance by more than 50%; whereas an atmospheric forcing with observed memory produces an SST variability comparable to observations. Approximately 70% of the extratropical atmospheric memory evidently originates from the tropical oceans, including the Niño3.4 region and the tropical North Atlantic, highlighting the significance of tropical-extratropical teleconnection in maintaining the observed SSTA variability in the extratropics.