Zhenming Ji

Due to the substantial domain gaps in Remote Sensing (RS) images that are characterized by variabilities such as location, wavelength, and sensor type, Remote Sensing Domain Generalization (RSDG) has emerged as a critical and valuable research frontier, focusing on developing models that generalize effectively across diverse scenarios. However, research in this area remains underexplored: (1) Current cross-domain methods primarily focus on Domain Adaptation (DA), which adapts models to predefined domains rather than to unseen ones; (2) Few studies target the RSDG issue, especially for semantic segmentation tasks. Existing related models are developed for specific unknown domains, struggling with issues of underfitting on other unseen scenarios; (3) Existing RS foundation models tend to prioritize in-domain performance over cross-domain generalization. To this end, we introduce the first vision foundation model for RSDG semantic segmentation, CrossEarth. CrossEarth demonstrates strong cross-domain generalization through a specially designed data-level Earth-Style Injection pipeline and a model-level Multi-Task Training pipeline. In addition, for the semantic segmentation task, we have curated an RSDG benchmark comprising 32 semantic segmentation scenarios across various regions, spectral bands, platforms, and climates, providing comprehensive evaluations of the generalizability of future RSDG models. Extensive experiments on this collection demonstrate the superiority of CrossEarth over existing state-of-the-art methods.

The Indochinese Peninsula experiences a dry season with extensive biomass burning peaking from March to April. As the monsoon arrives, rainfall significantly removes aerosols through deposition, ending the emission season. However, the biomass burning aerosols exert an influence on the atmospheric circulation prior to the monsoon onset. This study employed statistical methods and a regional atmosphere–chemistry model [WRF Model coupled with chemistry (WRF-Chem)] to investigate the delayed impact of biomass burning from the Indochinese Peninsula on the monsoon onset. The results indicate that the cold sea surface temperature anomaly caused by the aerosols during the emission season can be stored in the ocean and inhibit convective activities over the adjacent sea regions in the postemission season, suppressing the southwestward cross-equatorial flow over the southern Bay of Bengal. This suppression delays the westward extension and northward shift of the upper-level South Asian high pressure system, along with its divergence and subsidence effects, thereby postponing the breakdown and retreat of the subtropical high-pressure belt. Simultaneously, the cold sea temperature also suppresses the development of a warm pool in the southeastern Bay of Bengal, which is associated with the generation of a monsoon onset vortex. Consequently, the onset of the Bay of Bengal monsoon is delayed. Due to the decreasing delayed effects of aerosols over time and the counteractive warming from the accelerated abnormal anticyclonic circulation in the upper-level Bay of Bengal, which results in accelerated disappearance of the cold signal in sea surface, the delayed influence of aerosols diminishes gradually after the onset of the Bay of Bengal monsoon until it disappears. Significance Statement: This study explores the correlation between spring biomass burning in the Indochinese Peninsula and the onset of the Bay of Bengal and South China Sea monsoons. The research reveals that biomass burning during the emission season might postpone the initiation of the Bay of Bengal monsoon, while it has minimal influence on the South China Sea monsoon. By employing sensitivity experiments using models, the study elucidates the underlying mechanisms responsible for these observed effects.

Many countries around the world are rapidly advancing sustainable development (SD) through the exploitation of clean energy sources such as solar and wind energy, which are becoming the core of the sustainable energy transition. In recent years, the continuous advancement of Earth system models (ESMs) has facilitated numerous studies utilizing them to predict long-term and large-scale meteorological elements, consequently enabling forecasts of wind and solar energy. These forecasts provide critical guidance for formulating national renewable energy policies. Nevertheless, the current literature on ESMs predicting wind and solar energy lacks sufficient integration. Hence, to comprehend the focal points and future research prospects, we conducted this systematic review, employing four academic search tools to comprehensively analyze the relevant literature from the past five years. We summarized the general analytical process and compared the content and conclusions of the literature. The study reveals that future photovoltaic (PV) potential for electricity generation may increase in certain regions but decrease in others, while the global potential for concentrated solar power (CSP) may diminish, influenced by diverse factors and displaying significant regional disparities. In addition, wind resource trends vary in different regions, and forecasts exhibit considerable uncertainty. Therefore, many studies have corrected wind speeds prior to predicting wind energy. Subsequent research endeavors should concentrate on optimizing ESMs, investigating the impacts of technological innovation, and enhancing the prediction and analysis of extreme weather events.

Unsupervised Domain Adaptation (UDA) for semantic segmentation has received widespread attention for its ability to transfer knowledge from the source to target domains without a high demand for annotations. However, semantic segmentation under adverse conditions still poses significant challenges for autonomous driving, as bad weather observation data may introduce unforeseeable problems. Although previous UDA works are devoted to adverse scene tasks, their adaptation process is redundant. For instance, unlabeled snow scene training data is a must for the model to achieve fair segmentation performance in snowy scenarios. We propose calling this type of adaptation process the Single to Single (STS) strategy. Clearly, STS is time-consuming and may show weaknesses in some comprehensive scenes, such as a night scene of sleet. Motivated by the concept of Domain Generalization (DG), we propose the Single to All (STA) model. Unlike DG, which trains models on one or multiple source domains without target domains, the STA model is based on UDA and employs one source domain, one target domain, and one introduced domain to achieve generalization to all adverse conditions by training on a single-scene dataset. Specifically, the STA model is advantageous as it learns from the source domain, reserves the style factors via a Reservation domain, and adapts the unified factors by the Randomization module. An Output Space Refusion module is also further incorporated to strengthen STA. Our STA achieves state-of-the-art performance in the Foggy Driving benchmark and demonstrates great domain generalizability in all conditions of the ACDC and Foggy Zurich benchmarks.

Ice records provide a qualitative rather than a quantitative indication of the trend of climate change. Using the bulk aerodynamic method and degree day model, this study quantified ice mass loss attributable to sublimation/evaporation (S/E) and meltwater on the basis of integrated observations (1960–2006) of glacier-related and atmospheric variables in the northeastern Tibetan Plateau. During 1961–2005, the average annual mass loss in the ice core was 95.33 ± 20.56 mm w.e. (minimum: 78.97 mm w.e. in 1967, maximum: 146.67 mm w.e. in 2001), while the average ratio of the revised annual ice accumulation was 21.2 ± 7.7% (minimum: 11.0% in 1992, maximum 44.8% in 2000). A quantitative formula expressing the relationship between S/E and air temperature at the monthly scale was established, which could be extended to estimation of S/E changes of other glaciers in other regions. The elevation effect on alpine precipitation determined using revised ice accumulation and instrumental data was found remarkable. This work established a method for quantitative assessment of the temporal variation in ice core mass loss, and advanced the reconstruction of long-term precipitation at high elevations. Importantly, the formula established for reconstruction of S/E from temperature time series data could be used in other regions.