4.7 Article

Improved Perceptron of Subsurface Chlorophyll Maxima by a Deep Neural Network: A Case Study with BGC-Argo Float Data in the Northwestern Pacific Ocean

期刊

REMOTE SENSING
卷 14, 期 3, 页码 -

出版社

MDPI
DOI: 10.3390/rs14030632

关键词

subsurface chlorophyll maximum; deep neural network; Gaussian radial basis activation function; BGC-Argo; northwestern Pacific Ocean

资金

  1. Ministry of Science and Technology of the People's Republic of China [2019YFE0125000]
  2. National Nature Science Foundation of China-Shandong Joint Fund [U1906215]
  3. National Natural Science Foundation of China [41876032, 41890805]

向作者/读者索取更多资源

This study developed an improved deep neural network model that accurately retrieves the vertical profile of chlorophyll a concentration and associated characteristics from surface-ocean data. The model performs well in regions with low surface chlorophyll a and has been validated using observations in the northwestern Pacific Ocean. Additionally, the model's application to infer vertical chlorophyll a profiles from remote-sensing information was also validated.
Subsurface chlorophyll maxima (SCMs), commonly occurring beneath the surface mixed layer in coastal seas and open oceans, account for main changes in depth-integrated primary production and hence significantly contribute to the global carbon cycle. To fill the gap of previous methods (in situ measurement, remote sensing, and the extrapolating function based on surface-ocean data) for obtaining SCM characteristics (intensity, depth, and thickness), we developed an improved deep neural network (IDNN) model using a Gaussian radial basis activation function to retrieve the vertical profile of chlorophyll a concentration (Chl a) and associated SCM characteristics from surface-ocean data. The annually averaged SCM depth was further incorporated into the bias term and the Gaussian activation function to improve the estimation accuracy of the IDNN model. Based on the Biogeochemical-Argo (BGC-Argo) data acquired for three regions in the northwestern Pacific Ocean, vertical Chl a profiles produced by our improved DNN model using sea surface Chl a and sea surface temperature (SST) were in good agreement with the observations, especially in regions with low surface Chl a. Compared to other neural-network-based models with one hidden layer and a sigmoid activation function, the IDNN model retrieved vertical Chl a profiles well in more eutrophic subpolar regions. Furthermore, the application of the IDNN model to infer vertical Chl a profiles from remote-sensing information was validated in the northwestern Pacific Ocean.

作者

我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。

评论

主要评分

4.7
评分不足

次要评分

新颖性
-
重要性
-
科学严谨性
-
评价这篇论文

推荐

Article Economics

How does the deterrence of fraud affect firms' innovation outputs?

Kun Wang, Jianqiang Chen

Summary: This paper examines the impact of fraud deterrence on firms' innovation outputs. The study finds that fraud deterrence is positively related to firms' innovation outputs, especially when there is high information asymmetry or employee turnover.

APPLIED ECONOMICS LETTERS (2023)

Review Chemistry, Physical

Chemical-vapor-deposition-grown 2D transition metal dichalcogenides: A generalist model for engineering electrocatalytic hydrogen evolution

Yang Zhao, Jiazhao Huang, Jianqiang Chen, Youwen Liu, Tianyou Zhai

Summary: Two-dimensional transition metal dichalcogenides (TMDs) have shown excellent catalytic performance for hydrogen evolution and are considered suitable substitutes for commercial Pt-based catalysts. Chemical vapor deposition (CVD) is an important technique for synthesizing controllable and high-purity TMDs for electrocatalysis and electronic devices. Recent research progress in CVD-grown TMDs nanosheets for electrocatalytic hydrogen evolution, including synthesis factors and engineering strategies, has been presented in this review.

NANO RESEARCH (2023)

Article Environmental Sciences

Interannual variations in the nutrient cycle in the central Bohai Sea in response to anthropogenic inputs

Xiaokun Ding, Jie Shi, Xinyu Guo, Huiwang Gao, Sumei Liu, Wei Guo

Summary: There is growing concern about the impact of human activities on the coastal nutrient cycle, but the interannual variations in response to anthropogenic nutrient input have been rarely studied. This study used a hydrodynamic-ecological model to analyze the interannual variations in the nutrient cycle in the central Bohai Sea. The results showed increasing trends in nitrogen levels, but no noticeable interannual variations in phosphorus levels. Atmospheric nitrogen deposition played a significant role in the nitrogen cycle, and sedimentation processes were responsible for removing most of the anthropogenic nitrogen inputs.

CHEMOSPHERE (2023)

Article Oceanography

Biogeochemistry-ecosystem-social interactions on the Chinese continental margins

Su Mei Liu, Wen Liang, Xinyu Guo, Nian Wu, Wuchang Zhang, Xiujuan Shan, Hua-De Zhao, Juying Wang, Jin Huang

Summary: Chinese continental margins are undergoing significant environmental changes due to anthropogenic activities and climate change. This study focuses on the Bohai as a case study and examines the impacts of multiple stressors on ecosystems and social interactions. The Bohai ecosystem has been extensively modified, with changes in the coastline, nutrient concentrations, and fishery resources. It is important to implement strict policies for coastline protection and pollution control, as well as measures like fishing bans and stock enhancement to mitigate these impacts.

OCEANOLOGIA (2023)

Article Environmental Sciences

Interannual variation of coastal upwelling around Hainan Island

Junying Zhu, Quanyi Zhou, Qianqing Zhou, Xinxing Geng, Jie Shi, Xinyu Guo, Yang Yu, Ziwei Yang, Renfu Fan

Summary: Coastal upwelling around Hainan Island has a significant impact on the physical environment, ecological environment, and fishery resources of the surrounding waters. The study analyzed the interannual variations of sea surface temperature (SST) around Hainan Island based on satellite observations from 2003 to 2021, focusing on the upwelling regions in Qiongdong and Qiongxi. The results showed warming trends in the upwelling regions, and the SST variations were influenced by factors such as wind stress curl, stratification, and bottom cold water in the Gulf of Tonkin.

FRONTIERS IN MARINE SCIENCE (2023)

Article Geosciences, Multidisciplinary

Low Sea Surface Salinity Event of the Japan Sea During the Last Glacial Maximum

Junyong Zheng, Xinyu Guo, Haiyan Yang, Kailun Du, Xinyan Mao, Wensheng Jiang, Takuya Sagawa, Yasumasa Miyazawa, Sergey M. Varlamov, Ayako Abe-Ouchi, Wing-Le Chan

Summary: By using a simple box model, researchers reconstructed the sea surface salinity in the Japan Sea over the past 35 kyr and found that the sea experienced a low salinity event during the Last Glacial Maximum, with the minimum value occurring at 20 kyr BP. It was concluded that both the inflow transport and the sea surface freshwater flux determine the exact value and timing of minimum salinity.

PALEOCEANOGRAPHY AND PALEOCLIMATOLOGY (2023)

Article Computer Science, Artificial Intelligence

Enhancing discriminative appearance model for visual tracking

Xuedong He, Calvin Yu-Chian Chen

Summary: Most end-to-end discriminative trackers aim to improve performance by mining more representative target features, but it is challenging to obtain comprehensive features due to diverse challenges. To address this issue, we propose a novel feature enhancement module for richer and comprehensive feature representation, which is trainable end-to-end. Additionally, we use a metric learning method to devise a verifier for adaptive update of the sample memory and learning rate.

EXPERT SYSTEMS WITH APPLICATIONS (2023)

Article Chemistry, Physical

Geometric Interaction Graph Neural Network for Predicting Protein-Ligand Binding Affinities from 3D Structures (GIGN)

Ziduo Yang, Weihe Zhong, Qiujie Lv, Tiejun Dong, Calvin Yu -Chian Chen

Summary: This paper proposes a geometric interaction graph neural network (GIGN) that incorporates 3D structures and physical interactions for predicting protein-ligand binding affinities. GIGN achieves state-of-the-art performance on three external test sets, and the learned representations of protein-ligand complexes are biologically meaningful.

JOURNAL OF PHYSICAL CHEMISTRY LETTERS (2023)

Article Computer Science, Artificial Intelligence

Attention fusion and target-uncertain detection for discriminative tracking

Xuedong He, Calvin Yu-Chian Chen

Summary: This paper proposes a lightweight feature separation and fusion module to improve the performance of discriminative trackers. Additionally, a target uncertain detection technique is designed to address the problem of tracking model corruption. Through comprehensive experimental evaluations, the results demonstrate the excellent performance of the proposed methods on seven public benchmarks.

KNOWLEDGE-BASED SYSTEMS (2023)

Article Computer Science, Artificial Intelligence

DSIL-DDI: A Domain-Invariant Substructure Interaction Learning for Generalizable Drug-Drug Interaction Prediction

Zhenchao Tang, Guanxing Chen, Hualin Yang, Weihe Zhong, Calvin Yu-Chian Chen

Summary: This article introduces a pluggable substructure interaction module called DSIL-DDI, which can learn domain-invariant representations of DDI from the source domain. It improves the generalization and interpretability of DDI prediction modeling and provides valuable insights for OOD DDI predictions.

IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS (2023)

Article Computer Science, Artificial Intelligence

LDCNet: Lightweight dynamic convolution network for laparoscopic procedures image segmentation

Yiyang Yin, Shuangling Luo, Jun Zhou, Liang Kang, Calvin Yu-Chian Chen

Summary: Medical image segmentation is crucial for modern healthcare systems, especially in reducing surgical risks and planning treatments. Transanal total mesorectal excision (TaTME) has become an important method for treating colon and rectum cancers. Real-time instance segmentation during TaTME surgeries can assist surgeons in minimizing risks. However, the dynamic variations in TaTME images pose challenges for accurate instance segmentation.

NEURAL NETWORKS (2024)

Article Computer Science, Artificial Intelligence

Progressive network based on detail scaling and texture extraction: A more general framework for image deraining

Jiehui Huang, Zhenchao Tang, Xuedong He, Jun Zhou, Defeng Zhou, Calvin Yu-Chian Chen

Summary: This study proposes a progressive learning multi-scale feature blending model for image deraining tasks. The model utilizes detail dilation and texture extraction to improve the restoration of rainy images. Experimental results show that the model achieves near state-of-the-art performance in rain removal tasks and exhibits better rain removal realism.

NEUROCOMPUTING (2024)

Article Chemistry, Physical

Equivariant Flexible Modeling of the Protein-Ligand Binding Pose with Geometric Deep Learning

Tiejun Dong, Ziduo Yang, Jun Zhou, Calvin Yu-Chian Chen

Summary: This article introduces a novel geometric graph network called FlexPose, which is capable of directly modeling flexible protein-ligand complex structures. The model outperforms popular docking tools and recent deep learning methods, especially in tasks involving protein conformation changes. It also provides affinity estimation and model confidence for post-analysis.

JOURNAL OF CHEMICAL THEORY AND COMPUTATION (2023)

Article Automation & Control Systems

MFAE: Multilevel Feature Aggregation Enhanced Drug-Target Affinity Prediction for Drug Repurposing Against Colorectal Cancer

Guanxing Chen, Haohuai He, Calvin Yu-Chian Chen

Summary: This study constructs a drug-target affinity dataset for two novel therapeutic targets for CRC and designs a deep learning-based MFAE model. The model simulates complex drug-target interactions through hierarchical feature extraction and multilevel feature aggregation and enhancement. Evaluation demonstrates the model's superior predictive accuracy, and fine-tuning enhances its performance. Two potential repurposing candidates are identified through virtual screening.

ADVANCED INTELLIGENT SYSTEMS (2023)

Article Radiology, Nuclear Medicine & Medical Imaging

MRI-based automated machine learning model for preoperative identification of variant histology in muscle-invasive bladder carcinoma

Jingwen Huang, Guanxing Chen, Haiqing Liu, Wei Jiang, Siyao Mai, Lingli Zhang, Hong Zeng, Shaoxu Wu, Calvin Yu-Chian Chen, Zhuo Wu

Summary: This study developed a non-invasive automated machine learning (AutoML) model to differentiate urothelial carcinoma with squamous differentiation (UC w/SD) from pure urothelial carcinoma (UC) in patients with muscle-invasive bladder carcinoma (MIBC). The model showed robust differentiation capabilities in both training and test sets, making it a potentially useful tool for pre-operative diagnosis.

EUROPEAN RADIOLOGY (2023)

暂无数据