Decoding river pollution trends and their landscape determinants in an ecologically fragile karst basin using a machine learning model.

Karst watersheds accommodate high landscape complexity and are influenced by both human-induced and natural activity, which affects the formation and process of runoff, sediment connectivity and contaminant transport and alters natural hydrological and nutrient cycling. However, physical monitoring stations are costly and labor-intensive, which has confined the assessment of water quality impairments on spatial scale. The geographical characteristics of catchments are potential influencing factors of water quality, often overlooked in previous studies of highly heterogeneous karst landscape. To solve this problem, we developed a machining learning method and applied Extreme Gradient Boosting (XGBoost) to predict the spatial distribution of water quality in the world's most ecologically fragile karst watershed. We used the Shapley Addition interpretation (SHAP) to explain the potential determinants. Before this process, we first used the water quality damage index (WQI-DET) to evaluate the water quality impairment status and determined that CODMn, TN and TP were causing river water quality impairments in the WRB. Second, we selected 46 watershed features based on the three key processes (sources-mobilization-transport) which affect the temporal and spatial variation of river pollutants to predict water quality in unmonitored reaches and decipher the potential determinants of river impairments. The predicting range of CODMn spanned from 1.39 mg/L to 17.40 mg/L. The predictions of TP and TN ranged from 0.02 to 1.31 mg/L and 0.25-5.72 mg/L, respectively. In general, the XGBoost model performs well in predicting the concentration of water quality in the WRB. SHAP explained that pollutant levels may be driven by three factors: anthropogenic sources (agricultural pollution inputs), fragile soils (low organic carbon content and high soil permeability to water flow), and pollutant transport mechanisms (TWI, carbonate rocks). Our study provides key data to support decision-making for water quality restoration projects in the WRB and information to help bridge the science:policy gap.

Temporal responses of hydrochemical variables and dissolved Fe(II) to flooding at a lake riparian wetland under different vegetation revealing by high resolution DGT.

The interplay between hydrological and biogeochemical processes in riparian wetland was recognized to lead directly to the temporal variations of surface water quality. However, the effects of flooding and vegetation on the release and entrapment of heavy metals and nutrients in riparian wetland remain poorly understood. The study aimed at investigating the influences of flooding and vegetation on the hydrochemical and Fe-redox change in the soil porewater and shallow groundwater, in Poyang lake riparian wetland through hydrochemical monitoring and diffusive gradient technology (DGT). The hydrochemical profiles and results of PCA analysis on the temporal datasets both demonstrated that vegetation had significant influences on the hydrochemistry of rhizosphere depth zone (RDZ) and shallow groundwater depth zone (SGZ). The Ca, K, Na, Mg, Mn and DOC at RDZ of both plants showed significant increasing trends from pre-to post-flooding while were observed minor change at the SGZ. The extracted PC1-PC3 from PCA analysis suggested that mineral dissolution and fermentation were dominating processes that explained 64.1% of the hydrochemical variability under the wetting-drying cycle. The synchronous changes of Fe(II), SO42-, DOC and ORP were found to occur at the SGZ of Carex, implying the likely occurrence of Fe- and S- redox reactions. The Fe(II) DGT profiles evidenced the temporal iron reduction and oxidation occurring at the rhizosphere following the wetting-drying cycle, as also reflected by the high opposite Fe2+ and DO association through PCA analysis. The high resolution temporal-spatial Fe(II) distribution suggested also the interface between lake water and groundwater was relatively stable under flooding. These results highlight that the release of dissolved Fe(II) from the wetland rhizosphere driven by flood may result in the release of Fe-associated heavy metals from riparian wetland to surface water, and hence pose potential threats to the surface water quality. Thereby, the flow and flood should be properly controlled and vegetation effects need to be carefully considered in the water resources management of lake-floodplain system.

Historical development, impact mechanism and future trends of nitrogen footprint in Wuxi City, China.

Human activities have changed the biogeochemical cycle of nitrogen, leading to a large amount of reactive nitrogen (Nr) into the environment, aggravating a series of environmental problems, affecting human and ecosystem health. Cities are the core areas driving nitrogen cycling in terrestrial ecosystems, however, there are numerous influencing factors and their contributions are unclear. The nitrogen footprint is an important index to understand the impact of human activities on the environment, however, the calculation of urban nitrogen footprint needs a simplified and accurate system method. Here we use a nitrogen footprint calculation model at the urban system level based on system nitrogen balance, and a multi-factor extended STIRPAT (stochastic impact by regression on population, affluence, and technology) model suitable for analyzing the impact mechanism of nitrogen footprint to estimate nitrogen footprint of Wuxi City during 1990-2050. We find that: (1) from 1990 to 2020, the total nitrogen footprint of Wuxi City was in an increasing trend, but the per capita nitrogen footprint was in a decreasing trend. The per capita nitrogen footprint of 22.36 kg capita-1 in 2020 was at a lower level globally. (2) Nr discharge from fossil fuel combustion and Haber-Bosch nitrogen fixation accounted for the main proportion of nitrogen footprint. (3) Dietary choice (Ad), GDP per capita (Ag), urbanization rate (Au), population (P), and fossil energy productivity (Te) were the key factors contributing to the increase of the nitrogen footprint, which resulted in an annual increase of 1.39 %. While nitrogen footprint productivity (Tn), nitrogen use efficiency in crop farming (Tc), and nitrogen use efficiency in animal breeding (Ta) were the key inhibit factors that inhibit the increase of nitrogen footprint, and these factors slow down the annual growth rate of nitrogen footprint by 0.39 %. (4) The continuous growth of nitrogen footprint in the baseline and population growth scenarios will bring more environmental problems and greater environmental governance pressure to Wuxi City, while the sustainable scenario that includes comprehensive means such as economic adaptation and technological improvement is more in line with the requirements of high-quality development in China. Several mitigation measures are then proposed by considering Wuxi's realities from both key impact factors and potential for nitrogen footprint reduction in different scenarios, which can provide valuable policy insights to other cities, especially lakeside cities to mitigate nitrogen footprint.

Spatio-temporal patterns and drivers of CH4 and CO2 fluxes from rivers and lakes in highly urbanized areas.

Gaseous carbon exchange at the water-air interface of rivers and lakes is an essential process for regional and global carbon cycle assessments. Many studies have shown that rivers surrounding urban landscapes can be hotspots for greenhouse gas (GHG) emissions. Here we investigated the variability of diffusive GHG (methane [CH4] and carbon dioxide [CO2]) emissions from rivers in different landscapes (i.e., urban, agricultural and mixed) and from lakes in Suzhou, a highly urbanized region in eastern China. GHG emissions in the Suzhou metropolitan water network followed a typical seasonal pattern, with the highest fluxes in summer, and were primarily influenced by temperature and dissolved oxygen concentration. Surprisingly, lakes were emission hotspots, with mean CH4 and CO2 fluxes of 2.80 and 128.89 mg m-2 h-1, respectively, translating to a total CO2-equivalent flux of 0.21 g CO2-eq m-2 d-1. The global warming potential of urban and mixed rivers (0.19 g CO2-eq m-2 d-1) was comparable to that for lakes, but about twice the value for agricultural rivers (0.10 g CO2-eq m-2 d-1). Factors related to the high GHG emissions in lakes included hypoxic water conditions and an adequate nutrient supply. Riverine CH4 emissions were primarily associated with the concentrations of total dissolved solids (TDS), ammonia­nitrogen and chlorophyll a. CO2 emissions in rivers were mainly closely related to TDS, with suitable conditions allowing rapid organic matter decomposition. Compared with other types of rivers, urban rivers had more available organic matter and therefore higher CO2 emissions. Overall, this study emphasizes the need for a deeper understanding of the impact of GHG emissions from different water types on global warming in rapidly urbanizing regions. Flexible management measures are urgently needed to mitigate CO2 and CH4 emissions more effectively in the context of the shrinking gap between urban and rural areas with growing socio-economic development.

High-Performance Acceleration of 2-D and 3-D CNNs on FPGAs Using Static Block Floating Point.

Over the past few years, 2-D convolutional neural networks (CNNs) have demonstrated their great success in a wide range of 2-D computer vision applications, such as image classification and object detection. At the same time, 3-D CNNs, as a variant of 2-D CNNs, have shown their excellent ability to analyze 3-D data, such as video and geometric data. However, the heavy algorithmic complexity of 2-D and 3-D CNNs imposes a substantial overhead over the speed of these networks, which limits their deployment in real-life applications. Although various domain-specific accelerators have been proposed to address this challenge, most of them only focus on accelerating 2-D CNNs, without considering their computational efficiency on 3-D CNNs. In this article, we propose a unified hardware architecture to accelerate both 2-D and 3-D CNNs with high hardware efficiency. Our experiments demonstrate that the proposed accelerator can achieve up to 92.4% and 85.2% multiply-accumulate efficiency on 2-D and 3-D CNNs, respectively. To improve the hardware performance, we propose a hardware-friendly quantization approach called static block floating point (BFP), which eliminates the frequent representation conversions required in traditional dynamic BFP arithmetic. Comparing with the integer linear quantization using zero-point, the static BFP quantization can decrease the logic resource consumption of the convolutional kernel design by nearly 50% on a field-programmable gate array (FPGA). Without time-consuming retraining, the proposed static BFP quantization is able to quantize the precision to 8-bit mantissa with negligible accuracy loss. As different CNNs on our reconfigurable system require different hardware and software parameters to achieve optimal hardware performance and accuracy, we also propose an automatic tool for parameter optimization. Based on our hardware design and optimization, we demonstrate that the proposed accelerator can achieve 3.8-5.6 times higher energy efficiency than graphics processing unit (GPU) implementation. Comparing with the state-of-the-art FPGA-based accelerators, our design achieves higher generality and up to 1.4-2.2 times higher resource efficiency on both 2-D and 3-D CNNs.

Design of Fully Spectral CNNs for Efficient FPGA-Based Acceleration.

Computing convolutional layers in the frequency domain using fast Fourier transformation (FFT) has been demonstrated to be effective in reducing the computational complexity of convolutional neural networks (CNNs). Nevertheless, the main challenge of this approach lies in the frequent and repeated transformations between the spatial and frequency domains due to the absence of nonlinear functions in the spectral domain, as such it makes the benefit less attractive for low-latency inference, especially on embedded platforms. To overcome the drawbacks in the existing FFT-based convolution, we propose a fully spectral CNN using a novel spectral-domain adaptive rectified linear unit (ReLU) layer, which completely removes the compute-intensive transformations between the spatial and frequency domains within the network. The proposed fully spectral CNNs maintain the nonlinearity of the spatial CNNs while taking into account the hardware efficiency. We then propose a deeply customized and compute-efficient hardware architecture to accelerate the fully spectral CNN inference on field programmable gate array (FPGA). Different hardware optimizations, such as spectral-domain intralayer and interlayer pipeline techniques, are introduced to further improve the performance of throughput. To achieve a load-balanced pipeline, a design space exploration (DSE) framework is proposed to optimize the resource allocation between hardware modules according to the resource constraints. On an Intel's Arria 10 SX160 FPGA, our optimized accelerator achieves a throughput of 204 Gop/s with 80% of compute efficiency. Compared with the state-of-the-art spatial and FFT-based implementations on the same device, our accelerator is 4 ×  âˆ¼  6.6 × and 3.0TEXPRESERVE3  âˆ¼  4.4 × faster while maintaining a similar level of accuracy across different benchmark datasets.

Toward Full-Stack Acceleration of Deep Convolutional Neural Networks on FPGAs.

Due to the huge success and rapid development of convolutional neural networks (CNNs), there is a growing demand for hardware accelerators that accommodate a variety of CNNs to improve their inference latency and energy efficiency, in order to enable their deployment in real-time applications. Among popular platforms, field-programmable gate arrays (FPGAs) have been widely adopted for CNN acceleration because of their capability to provide superior energy efficiency and low-latency processing, while supporting high reconfigurability, making them favorable for accelerating rapidly evolving CNN algorithms. This article introduces a highly customized streaming hardware architecture that focuses on improving the compute efficiency for streaming applications by providing full-stack acceleration of CNNs on FPGAs. The proposed accelerator maps most computational functions, that is, convolutional and deconvolutional layers into a singular unified module, and implements the residual and concatenative connections between the functions with high efficiency, to support the inference of mainstream CNNs with different topologies. This architecture is further optimized through exploiting different levels of parallelism, layer fusion, and fully leveraging digital signal processing blocks (DSPs). The proposed accelerator has been implemented on Intel's Arria 10 GX1150 hardware and evaluated with a wide range of benchmark models. The results demonstrate a high performance of over 1.3 TOP/s of throughput, up to 97% of compute [multiply-accumulate (MAC)] efficiency, which outperforms the state-of-the-art FPGA accelerators.

In silico analysis of glycosyltransferase 2 family genes in duckweed (Spirodela polyrhiza) and its role in salt stress tolerance.

Plant glycosyltransferase 2 (GT2) family genes are involved in plant abiotic stress tolerance. However, the roles of GT2 genes in the abiotic resistance in freshwater plants are largely unknown. We identified seven GT2 genes in duckweed, remarkably more than those in the genomes of Arabidopsis thaliana, Oryza sativa, Amborella trichopoda, Nymphaea tetragona, Persea americana, Zostera marina, and Ginkgo biloba, suggesting a significant expansion of this family in the duckweed genome. Phylogeny resolved the GT2 family into two major clades. Six duckweed genes formed an independent subclade in Clade I, and the other was clustered in Clade II. Gene structure and protein domain analysis showed that the lengths of the seven duckweed GT2 genes were varied, and the majority of GT2 genes harbored two conserved domains, PF04722.12 and PF00535.25. The expression of all Clade I duckweed GT2 genes was elevated at 0 h after salt treatment, suggesting a common role of these genes in rapid response to salt stress. The gene Sp01g00794 was highly expressed at 12 and 24 h after salt treatment, indicating its association with salt stress resilience. Overall, these results are essential for studies on the molecular mechanisms in stress response and resistance in aquatic plants.