PSLT: A Light-Weight Vision Transformer With Ladder Self-Attention and Progressive Shift.

Vision Transformer (ViT) has shown great potential for various visual tasks due to its ability to model long-range dependency. However, ViT requires a large amount of computing resource to compute the global self-attention. In this work, we propose a ladder self-attention block with multiple branches and a progressive shift mechanism to develop a light-weight transformer backbone that requires less computing resources (e.g., a relatively small number of parameters and FLOPs), termed Progressive Shift Ladder Transformer (PSLT). First, the ladder self-attention block reduces the computational cost by modelling local self-attention in each branch. In the meanwhile, the progressive shift mechanism is proposed to enlarge the receptive field in the ladder self-attention block by modelling diverse local self-attention for each branch and interacting among these branches. Second, the input feature of the ladder self-attention block is split equally along the channel dimension for each branch, which considerably reduces the computational cost in the ladder self-attention block (with nearly [Formula: see text] the amount of parameters and FLOPs), and the outputs of these branches are then collaborated by a pixel-adaptive fusion. Therefore, the ladder self-attention block with a relatively small number of parameters and FLOPs is capable of modelling long-range interactions. Based on the ladder self-attention block, PSLT performs well on several vision tasks, including image classification, objection detection and person re-identification. On the ImageNet-1 k dataset, PSLT achieves a top-1 accuracy of 79.9% with 9.2 M parameters and 1.9 G FLOPs, which is comparable to several existing models with more than 20 M parameters and 4 G FLOPs. Code is available at https://isee-ai.cn/wugaojie/PSLT.html.

Characteristics of surface water quality and stable isotopes in Bamen Bay watershed, Hainan Province, China.

Bamen Bay is located at the intersection of the Wenjiao River and Wenchang River in Hainan Province (China), where mangroves have been facing a threat of water quality deterioration. Therefore, it is imperative to study the characteristics of the surface water quality on a watershed scale. Water samples were collected three times from 36 monitoring sites from 2015 to 2016. It was found that nitrate was the main inorganic nitrogen form and all the surface water types were alkaline. Meanwhile, aquaculture water had high content of nitrogen, total phosphorus, chlorophyll a (Chl.a), total organic carbon (TOC), and chemical oxygen demand (COD). Significant spatial and temporal variations were found for most parameters. However, stable isotopes of δD and δ18O indicated that river water mainly originated from atmospheric precipitation and experienced strong evaporation. The water chemistry and isotopes of the Bamen Bay, mangroves, and aquaculture water were initially affected by the mixing of fresh water and seawater, followed by evaporation. The river and reservoir water chemistry were mainly controlled by water-rock interactions and cation exchange as deduced from the ionic relationships and Gibbs plots. These interactions involved the dissolution of calcite-, bicarbonate-, carbonate-, and calcium-containing minerals. Oxidized environments (river, reservoir, and Bamen Bay) were conducive for nitrification, while anaerobic conditions (mangrove and aquaculture water) were beneficial to the reduced nitrogen forms.

Riverine nutrient fluxes and environmental effects on China's estuaries.

An increase in riverine nutrient fluxes significantly influences the estuarine ecosystem. This study collected nutrient data in most of China's rivers from 1963 to 2015 to estimate the nutrient fluxes from major rivers and analyze interannual variability of nutrient fluxes and estuarine environmental effects. The results showed that the nutrient fluxes from the Yangtze River increased annually from 1963 to 2012. The trend of nutrient fluxes from the Yellow River was consistent with that from the Jiulong River, i.e., nutrient fluxes increased from 1998 to 2007 and then decreased. The areal nutrient fluxes from China's major rivers were higher than those from major world rivers, while the areal nutrient yield rates per capita were lower than those from major world rivers. We also found that China's estuaries were predominantly phosphorus-limited and slowly moving toward lower dissolved silica over dissolved inorganic nitrogen (DSi:DIN) ratios with time. Meanwhile, the nutrient limitation of phytoplankton growth in most of China's estuary systems was moving toward a higher incidence of phosphorus and silicon limitations as a result of increased DIN fluxes, and this would likely alter phytoplankton communities. Furthermore, the decreases in the DSi:DIN ratio and dissolved silica over dissolved inorganic phosphate (DSi:DIP) ratio, and the increases in both DIN and DIP fluxes, caused increased red tide blooms.

Systematic approach to evaluating environmental and ecological technologies for wastewater treatment.

Evaluating the performance of wastewater treatment represents a challenging and complex task as it usually involves engineering, environmental and economic (3E) factors. In this study, we developed an 3E triangle model to evaluate the performance of environmental technologies (i.e., anaerobic-anoxic-oxic reactors, oxidation ditches, and membrane bioreactors) and ecological technologies (i.e., stabilization ponds, constructed wetlands, and slow-rate systems) for wastewater treatment. A total of 17 key performance indicators, such as energy consumption, pollutant removal, global warming potential and wastewater treatment fees, were considered in the 3E triangle model. The results indicated that, in terms of engineering performance, both the membrane bioreactors and constructed wetlands were stable, effective and reliable during their operating periods. When the environmental impacts of wastewater treatment technologies were compared via a life cycle assessment, the ecological technologies showed superior performance, in terms of environmental impacts, especially for the global warming potential and eutrophication potential. In general, environmental technologies exhibited higher treatment fees and unit construction costs because of their large power consumption and equipment costs. In contrast, ecological technologies had higher unit land use due to their large area requirements and low treatment capacity. In overall, both the membrane bioreactors and constructed wetlands showed excellent overall performance in the 3E triangle model. Wastewater treatment plant are typical case studies for addressing the interactions of water and energy elements. Reducing energy consumption is a hotspot for the research field of membrane bioreactors, while constructed wetlands are continually improved and optimized to have broad applications for rural wastewater treatment.

Contrasting fine-scale genetic structure of two sympatric clonal plants in an alpine swampy meadow featured by tussocks.

Tussocks are unique vegetation structures in wetlands. Many tussock species mainly reproduce by clonal growth, resulting in genetically identical offspring distributed in various spatial patterns. These fine-scale patterns could influence mating patterns and thus the long-term evolution of wetland plants. Here, we contribute the first genetic and clonal structures of two key species in alpine wetlands on the Qinghai-Tibet Plateau, Kobresia tibetica and Blysmus sinocompressus, using > 5000 SNPs identified by 2b-RAD sequencing. The tussock-building species, K. tibetica, has a phalanx (clumping) growth form, but different genets could co-occur within the tussocks, indicating that it is not proper to treat a tussock as one genetic individual. Phalanx growth does not necessarily lead to increased inbreeding in K. tibetica. B. sinocompressus has a guerilla (spreading) growth form, with the largest detected clone size being 18.32 m, but genets at the local scale tend to be inbred offspring. Our results highlight that the combination of clone expansion and seedling recruitment facilitates the contemporary advantage of B. sinocompressus, but its evolutionary potential is limited by the input genetic load of the original genets. The tussocks of K. tibetica are more diverse and a valuable genetic legacy of former well-developed wet meadows, and they are worthy of conservation attention.

Assessment of nitrogen reduction by constructed wetland based on InVEST: A case study of the Jiulong River Watershed, China.

The Jiulong River watershed (JRW) in southeast China includes livestock breeding and agriculture, leading to large amounts of non-point source pollution. Nitrogen (N) reductions were simulated and mapped using the Integrated Valuation of Ecosystem Services and Tradeoffs (InVEST) under scenarios that were built considered both constructed wetlands (CWs) and climate change, which are not common in the literature on ecosystem services assessments. The results showed that the amount of N exported from non-point sources within the JRW was 12,569 t·yr-1. The areal N load was relatively higher in the north, while more N exported in the southeast. Constructed riparian wetlands can intercept and reduce the N loads that enter water bodies, but climate change may be a factor driving the deterioration of water quality. The methodology can be generalized to reduce other contaminants, and provides a tool for decision-makers to weigh the costs and benefits of urbanization and conservation.

Decadal changes in nutrient fluxes and environmental effects in the Jiulong River Estuary.

Estuaries are areas of both freshwater and seawater that are partially enclosed with contact to the open sea and a flow of fresh water. Although the Jiulong River Estuary has a relatively small catchment, this area was found to exhibit high nutrient fluxes. The nutrient fluxes showed obvious fluctuations for different years. The Jiulong River Estuary was predominantly P-limited, and was slowly moving towards higher DIN:DIP and DSi:DIP ratios as the nitrate concentrations increased. The high nutrient fluxes into the estuary may affect estuarine ecosystems by the alteration of DO concentrations in bottom waters, causing harm to benthic fauna due to a lack of oxygen, triggering algal blooms. Additionally, the Jiulong River Estuary was slowly moving towards lower DSi:DIN and DSi:DIP ratios along with the change of time scales, which caused nutrient limitation of phytoplankton growth as P and Si levels decreased and became more limiting.