Lingfeng Wan

Two of the challenges for paleoclimate data assimilation (PDA) are the extreme sparsity of proxy observations and the non‐Gaussian nature of background errors in the climate system. In this study, we develop a latent‐space variational data assimilation framework based on variational autoencoders (VAEs) and latent‐space minimization (VAE‐Var), and apply it to assimilating sparse air and sea temperature proxies. Employing a VAE‐based encoder–decoder to learn a data‐driven latent prior that renders non‐Gaussian background errors, VAE‐Var addresses the challenge of non‐Gaussian background errors through an adjustable Gaussian constraint, relaxing the Gaussian‐error assumption underpinning traditional variational schemes. Using surface air temperature and sea surface temperature as targets, we systematically evaluate “reconstruction” skills across varying levels of observational sparsity, comparing VAE‐Var with conventional approaches. Diagnostics of analysis increments demonstrate that VAE‐Var yields improved “reconstruction” skill and more physically plausible analyses under sparse observations, highlighting its merits in non‐Gaussian PDA for high‐dimensional reconstruction applications.

For a long time, the 4.2 ka event was considered sufficiently significant to mark the onset of the Meghalayan Age, the latest stage of the Holocene. However, the spatial distribution and temporal evolutions of precipitation changes remain controversial in many regions, especially in monsoon Asia. In this study, combining the high‐resolution speleothem calcite oxygen isotope ratio value (δ ¹⁸ O _c ) records and the transient simulation of an isotope enabled earth system model, we reconstruct the spatiotemporal evolution of δ ¹⁸ O _c and precipitation in the Asian monsoon region during the 4.2 ka event (4.3–3.9 ka BP) with 10‐year temporal resolution and 2.5° × 1.9° spatial resolution, using an offline ensemble Kalman filter. Based on these reconstructed results, we analyze the characteristics of the 4.2 ka event over monsoon Asia from two aspects. On the centennial scale, the event shows a drought pattern in northern China and South Asia, and a wet pattern in southern China. On the multi‐decadal scale, frequent wet‐dry excursions occur in the region. Notably, both the centennial‐scale changes and multi‐decadal excursions largely fall within the range of Holocene natural variability. These results challenge the conventional view of the 4.2 ka event as a global megadrought and instead suggest that, even within monsoon Asia, it was expressed as a spatially heterogeneous hydroclimatic anomaly rather than a spatially uniform or exceptionally extreme event.

The climatic interpretations of speleothem δ ¹⁸ O _c records over Asian summer monsoon (ASM) region have been controversial. Here, combining speleothem δ ¹⁸ O _c records with the first transient Holocene simulation of an isotope‐enabled earth system model, we reconstruct a dynamically consistent spatio‐temporal evolution of the ASM precipitation using paleoclimate data assimilation. Our reconstruction finds a new mega‐tripolar pattern of precipitation variations accompanying a continental‐wide enrichment of δ ¹⁸ O _c from the early to late Holocene over the entire ASM continental region. Precipitation peaked in the early Holocene over South Asia, northern China and southern China, but in the late Holocene over central China. This mega‐tripolar pattern of precipitation variations can help resolve the long‐standing controversy about the timing of Holocene ASM precipitation peaks reconstructed from various proxy records.

Proxy reconstructions suggest that increasing global mean sea surface temperature (GMSST) during the last deglaciation was accompanied by a comparable or greater increase in global mean ocean temperature (GMOT), corresponding to a large heat storage efficiency (HSE; ∆GMOT/∆GMSST). An increased GMOT is commonly attributed to surface warming at sites of deepwater formation, but winter sea ice covered much of these source areas during the last deglaciation, which would imply an HSE much less than 1. Here, we use climate model simulations and proxy-based reconstructions of ocean temperature changes to show that an increased deglacial HSE is achieved by warming of intermediate-depth waters forced by mid-latitude surface warming in response to greenhouse gas and ice sheet forcing as well as by reduced Atlantic meridional overturning circulation associated with meltwater forcing. These results, which highlight the role of surface warming and oceanic circulation changes, have implications for our understanding of long-term ocean heat storage change.