Environmental Datasets Research Articles

European marine omics biodiversity observation network: a strategic outline for the implementation of omics approaches in ocean observation

Marine ecosystems, ranging from coastal seas and wetlands to the open ocean, accommodate a wealth of biological diversity from small microorganisms to large mammals. This biodiversity and its associated ecosystem function occurs across complex spatial and temporal scales and is not yet fully understood. Given the wide range of external pressures on the marine environment, this knowledge is crucial for enabling effective conservation measures and defining the limits of sustainable use. The development and application of omics-based approaches to biodiversity research has helped overcome hurdles, such as allowing the previously hidden community of microbial life to be identified, thereby enabling a holistic view of an entire ecosystem’s biodiversity and functioning. The potential of omics-based approaches for marine ecosystems observation is enormous and their added value to ecosystem monitoring, management, and conservation is widely acknowledged. Despite these encouraging prospects, most omics-based studies are short-termed and typically cover only small spatial scales which therefore fail to include the full spatio-temporal complexity and dynamics of the system. To date, few attempts have been made to establish standardised, coordinated, broad scaled, and long-term omics observation networks. Here we outline the creation of an omics-based marine observation network at the European scale, the European Marine Omics Biodiversity Observation Network (EMO BON). We illustrate how linking multiple existing individual observation efforts increases the observational power in large-scale assessments of status and change in biodiversity in the oceans. Such large-scale observation efforts have the added value of cross-border cooperation, are characterised by shared costs through economies of scale, and produce structured, comparable data. The key components required to compile reference environmental datasets and how these should be linked are major challenges that we address.

Standardized Data to Improve Understanding and Modeling of Soil Nitrogen at Continental Scale

AbstractNitrogen (N) is a key limiting nutrient in terrestrial ecosystems, but there remain critical gaps in our ability to predict and model controls on soil N cycling. This may be in part due to lack of standardized sampling across broad spatial–temporal scales. Here, we introduce a continentally distributed, publicly available data set collected by the National Ecological Observatory Network (NEON) that can help fill these gaps. First, we detail the sampling design and methods used to collect and analyze soil inorganic N pool and net flux rate data from 47 terrestrial sites. We address methodological challenges in generating a standardized data set, even for a network using uniform protocols. Then, we evaluate sources of variation within the sampling design and compare measured net N mineralization to simulated fluxes from the Community Earth System Model 2 (CESM2). We observed wide spatiotemporal variation in inorganic N pool sizes and net transformation rates. Site explained the most variation in NEON’s stratified sampling design, followed by plots within sites. Organic horizons had larger pools and net N transformation rates than mineral horizons on a sample weight basis. The majority of sites showed some degree of seasonality in N dynamics, but overall these temporal patterns were not matched by CESM2, leading to poor correspondence between observed and modeled data. Looking forward, these data can reveal new insights into controls on soil N cycling, especially in the context of other environmental data sets provided by NEON, and should be leveraged to improve predictive modeling of the soil N cycle.

Data quality control considerations in multivariate environmental monitoring: experience of the French coastal network SOMLIT

IntroductionWhile crucial to ensuring the production of accurate and high-quality data—and to avoid erroneous conclusions—data quality control (QC) in environmental monitoring datasets is still poorly documented.MethodsWith a focus on annual inter-laboratory comparison (ILC) exercises performed in the context of the French coastal monitoring SOMLIT network, we share here a pragmatic approach to QC, which allows the calculation of systematic and random errors, measurement uncertainty, and individual performance. After an overview of the different QC actions applied to fulfill requirements for quality and competence, we report equipment, accommodation, design of the ILC exercises, and statistical methodology specially adapted to small environmental networks (<20 laboratories) and multivariate datasets. Finally, the expanded uncertainty of measurement for 20 environmental variables routinely measured by SOMLIT from discrete sampling—including Essential Ocean Variables—is provided.Results, Discussion, ConclusionThe examination of the temporal variations (2001–2021) in the repeatability, reproducibility, and trueness of the SOMLIT network over time confirms the essential role of ILC exercises as a tool for the continuous improvement of data quality in environmental monitoring datasets.

Autonomous Path Finding and Obstacle Avoidance Method for Unmanned Construction Machinery

The working environment of construction machinery is harsh, and some operations are highly repetitive. The realization of intelligent construction machinery helps to improve economic efficiency and promote industrial development. Construction machinery is different from ordinary passenger vehicles. Aiming at the fact that the existing environmental perception data set cannot be directly applied to construction machinery, this paper establishes the corresponding data set in combination with the specific working conditions of construction machinery and carries out training based on the PointPillars network to realize the environmental perception function applicable to the working conditions of construction machinery. Most construction machinery runs on unstructured roads, and the existing passenger vehicle path planning algorithm is not applicable to construction machinery. Based on this, this paper uses a hybrid A* algorithm to achieve path planning that meets the kinematics of construction machinery and realizes real-time obstacle detection and avoidance. At the same time, this paper combines environmental perception with a path planning algorithm to provide a method of autonomous path finding and obstacle avoidance for construction machinery. Based on the improved pure pursuit algorithm, the high-precision motion control and established trajectory tracking of construction machinery are realized, which lays a certain foundation for the follow-up research and development of related intelligent technologies of construction machinery.

Voice Detection Using Convolutional Neural Network

The article presents an approach , methodology , the software system based on a machine learning technologies for convolutional neural network and its use for voice ( cough ) recognition . Tasks of article are receiving evaluating a voice detection system with deep learning , the use of a convolutional neural network and Python language for patients with cough . The convolutional neural network has been developed , trained and tested using various datasets and Python libraries . Unlike the existing modern works related to this area , proposed system was evaluated using a real set of environmental sound data , and not only on filtered or separated voice audio tracks . The final compiled model showed a relatively high average accuracy of 85.37 %. Thus , the system is able to detect the sound of a voice in a crowded public place , and there is no need for a sound separation phase for pre-processing , as other modern systems require . Several volunteers recorded their voice sounds using microphones of their smartphones , and it was guaranteed that they would test their voices in public places to make noise , in addition to some audio files that were uploaded online . The results showed an average recognition accuracy – of 85.37 %, a minimum accuracy – of 78.8 % and a record – of 91.9 %.

Clustered Distributed Learning Exploiting Node Centrality and Residual Energy (CINE) in WSNs

With the explosion of big data, the implementation of distributed machine learning mechanisms in wireless sensor networks (WSNs) is becoming required for reducing the number of data traveling throughout the network and for identifying anomalies promptly and reliably. In WSNs, the above need has to be considered along with the limited energy and processing resources available at the nodes. In this paper, we tackle the resulting complex problem by designing a multi-criteria protocol CINE that stands for “Clustered distributed learnIng exploiting Node centrality and residual Energy” for distributed learning in WSNs. More specifically, considering the energy and processing capabilities of nodes, we design a scheme that assumes that nodes are partitioned in clusters and selects a central node in each cluster, called cluster head (CH), that executes the training of the machine learning (ML) model for all the other nodes in the cluster, called cluster members (CMs). In fact, CMs are responsible for executing the inference only. Since the CH role requires the consumption of more resources, the proposed scheme rotates the CH role among all nodes in the cluster. The protocol has been simulated and tested using real environmental data sets.

A hierarchical birdsong feature extraction architecture combining static and dynamic modeling

To conserve bird biodiversity and monitor the distribution of species in the region, it is of tremendous necessity to identify birds by their songs and explore the rich ecological information birdsong contains. The audios recorded in the monitoring area generally have complex background noise, the characteristics of the song are not prominent and the biological spectrum information is not comprehensive, which brings some challenges to the identification of birds. This study proposes a hierarchical birdsong feature extraction architecture combining dynamic and static modeling to cope with complex environments as a modeling context. Firstly, six common speech features were extracted for the characteristics of birdsong. The Pearson correlation coefficient is then used to analyze the correlations between birdsong and human speech, examining the correlations between each feature in the presence and absence of environmental noise interference. Combined with the scatter plot matrix analysis, we conclude that Mel Frequency Cepstral Coefficient (MFCC) is more suitable comparing with other features when dealing with birdsong and can superiorly cope with a complex background noise. Secondly, a feature extraction architecture is built, which integrates static and dynamic modeling to fully explore the contextual relationship, to solve the problem of ignoring the internal structure information of the patch and losing some spatial information in the Transformer-type model. Finally, a hierarchical refinement module is designed to help extract more detailed features, as well as to optimize the computational cost of the Transformer-type model that requires many training data and has high complexity. The performance of the model can be detected with 93.67 % accuracy on the self-built birdsong dataset, 95.19 % accuracy on the public birdsong dataset Birdsdata and 97.02 % accuracy on the public environmental dataset UrbanSound8k.

A novel anomaly score based on kernel density fluctuation factor for improving the local and clustered anomalies detection of isolation forests

Isolation Forests (IF) and improved algorithms about anomaly scores are commonly used to detect global, local, or clustered anomalies. However, these algorithms are limited in considering all three anomalies simultaneously. To solve this problem, we propose a novel detection algorithm called KDF-IF. The KDF-IF algorithm redesigns the calculation of the anomaly score based on the Kernel Density Fluctuation (KDF) factor. By focusing on the density fluctuation of each instance neighborhood, the KDF draws the global, local, and clustered features for detecting the three types of anomalies. With the support of the Isolation Trees, the distance between instances in the traditional KDF is replaced by the path length to reduce the computational cost. Simulation results denote that the KDF-IF performance for three types of anomalies is significantly better than IF under the UCI datasets. Besides, the KDF-IF is used to effectively detect anomalies in the environmental dataset of intelligent networks, extending its application range.

Outlier Detection Algorithms for Open Environments

The high dimensionality and massive amount of data in open environments make the existing low-dimensional outlier detection methods time-consuming. The support vector machine (SVM) is a commonly used outlier detection method. However, the SVM still faces the problem of difficulty in obtaining optimal parameters quickly and effectively, resulting in low detection efficiency, poor stability, and difficulty in applying to open environment datasets. In order to improve the efficiency and stability of outlier detection, this paper proposes an improved sparrow search algorithm and uses it to optimize SVM parameters. First, the traditional sparrow search algorithm is improved by using improved backtracking learning and variable logarithmic spirals. Then, the improved sparrow search method is used to optimize SVM parameters, and the optimized support vector machine is applied to the field of outlier detection. Simulation experimental results show that the proposed method is significantly better than the compared classification algorithms in multiple evaluation indicators, with better detection efficiency, stability, and generalization ability.

Intelligent Parking Control Method Based on Multi-Source Sensory Information Fusion and End-to-End Deep Learning

To address the challenges of inefficient intelligent parking performance and reduced efficiency in complex environments, this study presents an end-to-end intelligent parking control model based on a Convolutional Neural Network–Long Short-Term Memory (CNN-LSTM) architecture incorporating multi-source sensory information fusion to improve decision-making and adaptability. The model can produce real-time intelligent parking control decisions by extracting spatiotemporal features, including comprehensive 360-degree panoramic images and ultrasonic sensor distance measurements. To enhance the coverage of real-world environments in the dataset, a data collection platform was developed, leveraging the PreScan simulation platform in conjunction with actual parking environments. Consequently, a comprehensive parking environment dataset comprising various types was constructed. A deep learning model was devised to ameliorate horizontal and vertical control in intelligent parking systems, integrating Convolutional Neural Networks and Long Short-Term Memory in a parallel configuration. By meticulously accounting for parking environment characteristics, sliding window parameters were optimized, and transfer learning was employed for secondary model training to fortify the prediction accuracy. To ascertain the system’s robustness, simulation tests were performed. The ultimate results from the actual environment experiment revealed that the end-to-end intelligent parking model substantially surpassed the existing approaches, bolstering parking efficiency and effectiveness in complex contexts.

Neighborhood Fast-Food Environments and Hypertension in Canadian Adults

Hypertension is a leading cause of cardiovascular disease and premature death worldwide. Neighbourhoods characterized by a high proportion of fast-food outlets may also contribute to hypertension in residents; however, limited research has explored these associations. This cross-sectional study assessed associations between neighbourhood fast-food environments, measured hypertension, and self-reported hypertension. Data from 10,700 adults living in urban areas was obtained from six Canadian Health Measures Survey cycles (2007-2019). Each participant's blood pressure was measured at a mobile clinic six times. Measured hypertension was defined as having an average SBP of ≥140, a DBP ≥90 mm Hg or being on blood pressure-lowering medication. Participants were also asked if they had been diagnosed with high blood pressure or if they take blood pressure-lowering medication (i.e., self-reported hypertension). The proportion of fast-food outlets relative to the sum of fast-food outlets and full-service restaurants in each participant's neighbourhood was obtained from the Canadian Food Environment Dataset, and analyses were conducted in 2022. The mean proportion of fast-food outlets was 23.3% (standard deviation (SD): 26.8%). A one SD increase in the proportion of fast-food outlets was associated with higher odds of measured hypertension in the full sample (OR=1.17, 95% CI 1.05, 1.31) and in sex-specific models (women: OR=1.14, 95% CI 1.01, 1.29; men: OR=1.21, 95% CI 1.03, 1.43). Associations between the proportion of fast-food outlets and self-reported hypertension were inconclusive. Findings suggest that reducing the proportion of fast-food restaurants in neighbourhoods may be a factor that could help reduce hypertension rates.

On the training sample size and classification performance: An experimental evaluation in seismic facies classification

Machine learning algorithms (MLAs) perform better when enough high-quality training data is provided. However, a lack of training data is frequent in seismic facies classification and many other supervised learning applications. Data labeling for seismic facies classification is time-consuming and requires considerable effort from the domain knowledge expert. This study investigates the effect of training data size on the performance of three popular supervised MLAs used for seismic facies classification. We labeled slices from two seismic datasets of diverse geologic environments and varying classification complexity. AN Field in Malay Basin represents a simple classification problem with three classes, whereas a more complex six classes classification is defined in the Dangerous Grounds (DG) dataset offshore Sabah. The labeled data were constantly reduced by half, resulting in eight training subsets of varying sizes. We trained and evaluated support vector machine (SVM), random forest (RF), and neural network (NN) models using a 10-fold cross-validation (CV) procedure. Performance metrics were computed to study the change in performance in response to the training data size. The experimental results show that, for the DG dataset, where the classification is complex due to the heterogeneous geology and a more number of classes, the larger the training subset, the better the classification performance. Nevertheless, for the simple classification scenario of the AN dataset, the classifiers reached a performance plateau when trained on limited samples. We found that the NN model is the best performer on large datasets. The RF classifier performed well in both datasets. It proved to be robust when trained on limited samples of the DG data. The SVM performed the best where there was a clear margin of separation between the defined classes (the AN data). In contrast, it performed poorly on the DG data and exhibited a performance decline on the AN large subsets.

A Shallow System Prototype for Violent Action Detection in Italian Public Schools

This paper presents a novel low-cost integrated system prototype, called School Violence Detection system (SVD), based on a 2D Convolutional Neural Network (CNN). It is used for classifying and identifying automatically violent actions in educational environments based on shallow cost hardware. Moreover, the paper fills the gap of real datasets in educational environments by proposing a new one, called Daily School Break dataset (DSB), containing original videos recorded in an Italian high school yard. The proposed CNN has been pre-trained with an ImageNet model and a transfer learning approach. To extend its capabilities, the DSB was enriched with online images representing students in school environments. Experimental results analyze the classification performances of the SVD and investigate how it performs through the proposed DSB dataset. The SVD, which achieves a recognition accuracy of 95%, is considered computably efficient and low-cost. It could be adapted to other scenarios such as school arenas, gyms, playgrounds, etc.

Statistical analysis of the effect of lead-bismuth eutectic on fatigue resistance of 316L

Environments like liquid lead-bismuth eutectic (LBE) or air can have an effect on the fatigue resistance of nuclear materials such as ferritic-martensitic and austenitic stainless steels if compared to vacuum. This paper provides a quantitative assessment, supported by statistical tools, of the effects of LBE and air on the fatigue life. Low cycle fatigue tests were performed on cylindrical samples of 316L in three environments: air, vacuum, and LBE (oxygen concentration <10–5 wt% and temperature 350 °C), and the experimental data was fitted to obtain life prediction models. The independent modelling of three separate environment datasets via the Langer equation showed that the slope of the fatigue curve is comparable among the three environments. In all models the sample standard deviation was found to be of the same order, suggesting that variations in microstructure and composition of 316L among samples have a similar impact in all three environments. A simplified model was then proposed considering vacuum as the reference condition, and assessing the effect of air and LBE as a quotient of lifetime in vacuum and air or lead-bismuth eutectic respectively. Analysis of the fracture surfaces indicated the existence of an LBE-steel interaction which seemed independent from oxygen content. For both air and LBE environments, fatigue life was found to be reduced by a factor of around 3 when compared to vacuum. Statistical analysis revealed that the difference in life reduction due to air and LBE is not significant.

Cross‐class pest and disease vegetation detection based on small sample registration

AbstractThis paper introduces few‐shot anomaly detection (FSAD), a practical and less anomaly detection (AD) method, which can provide a limited number of normal images for each class during training. So far, studies on FSAD have been carried out according to each model, and there is no discussion of commonalities between different types. Depending on how people detect unusual lies, the problematic images are compared to the normal ones. The image alignment method based on different classifications is used to train the target detection model independent of classification, and performed ablation experiments on the pest and disease datasets in different environments for verification. This is the first time the FSAD method has been used to train a single scalable model without the need to train new classifications or adjust parameters. The experimental results show that the application of AUC based on vegetation disease data set and vegetation pest data set in FSAD algorithm is improved by 19.5% compared with the existing algorithm.

Environmental viromes reveal the global distribution signatures of deep-sea DNA viruses

IntroductionViruses are abundant and ecologically significant in marine ecosystems. However, the virome of deep-sea sediments is not extensively investigated. ObjectivesTo explore the distribution pattern of deep-sea viruses on a global scale, the viromes of DNA viruses isolated from 138 sediments of 5 deep-sea ecosystems were characterized. MethodsThe viral particles were purified from each sediment sample. Then the viral DNAs were extracted and subjected to viral metagenomic analysis. ResultsHere, we constructed a global deep-sea environmental virome dataset by analyzing the viral DNA of 138 sediment samples. A total of 347,737 viral operational taxonomic units (vOTUs) were identified, of which 84.94% were hitherto unknown, indicating that deep sea was a reservoir of novel DNA viruses. Furthermore, circular viral genome analysis revealed 98,581 complete genomes. The classified vOTUs included eukaryotic (44.55%) and prokaryotic (25.75%) viruses, and were taxonomically assigned to 63 viral families. The composition and abundance of the deep-sea sediment viromes were dependent on the deep-sea ecosystem as opposed to geographical region. Further analysis revealed that the viral community differentiation in different deep-sea ecosystems was driven by the virus-mediated energy metabolism. ConclusionOur findings showed that deep-sea ecosystems are a reservoir of novel DNA viruses and the viral community is shaped by the environmental characteristics of deep-sea ecosystems, thus presenting critical information for determining the ecological significance of viruses in global deep-sea ecosystems.

A guide and tools for selecting and accessing microclimate data for mechanistic niche modeling

AbstractMost ecological analyses and forecasts use weather station data or coarse interpolated, gridded air temperature data. Yet, these products often poorly capture the microclimates experienced by organisms that respond to fine‐scale spatial and temporal environmental variation near the surface. Sources of historic and projected future data with finer spatial and temporal resolution are proliferating. We qualitatively and quantitatively review and evaluate the available data on three core issues central to microclimate modeling: the quality of the input environmental data, the ability of algorithms to capture microclimatic processes given environmental forcing data, and how best to access microclimatic data. We show how differences between observed environmental conditions and those estimated using environmental forcing data, microclimate algorithms, and precomputed microclimate datasets can be substantial depending on the variable, location, and season. The choice of environmental dataset to parameterize biophysical models has ramifications for biological estimates, such as the duration of potential activity and incidence of thermal stress. New data sources offering high temporal and spatial resolution correspond well to observational data and have the potential to revolutionize understanding of the ecological implications of microclimate variability. We provide resources to help users select and access appropriate environmental data for biological applications, including users' guides and interactive visualization, to better infer how organisms experience climate variability and change.

Reducing the complexity of high-dimensional environmental data: An analytical framework using LASSO with considerations of confounding for statistical inference.

Frameworks for selecting exposures in high-dimensional environmental datasets, while considering confounding, are lacking. We present a two-step approach for exposure selection with subsequent confounder adjustment for statistical inference. We measured cognitive ability in 338 children using the Woodcock-Muñoz General Intellectual Ability (GIA) score, and potential associated features across several environmental domains. Initially, 111 variables theoretically associated with GIA score were introduced into a Least Absolute Shrinkage and Selection Operator (LASSO) in a 50% feature selection subsample. Effect estimates for selected features were subsequently modeled in linear regressions in a 50% inference (hold out) subsample, first adjusting for sex and age and later for covariates selected via directed acyclic graphs (DAGs). All models were adjusted for clustering by school. Of the 15 LASSO selected variables, eleven were not associated with GIA score following our inference modeling approach. Four variables were associated with GIA scores, including: serum ferritin adjusted for inflammation (inversely), mother's IQ (positively), father's education (positively), and hours per day the child works on homework (positively). Serum ferritin was not in the expected direction. Our two-step approach moves high-dimensional feature selection a step further by incorporating DAG-based confounder adjustment for statistical inference.

Iterative mapping of marine ecosystems for spatial status assessment, prioritization, and decision support

South Africa has taken an iterative approach to marine ecosystem mapping over 18 years that has provided a valuable foundation for ecosystem assessment, planning and decision-making, supporting improved ecosystem-based management and protection. Iterative progress has been made in overcoming challenges faced by developing countries, especially in the inaccessible marine realm. Our aim is to report on the approach to produce and improve a national marine ecosystem map to guide other countries facing similar challenges, and to illustrate the impact of even the simplest ecosystem map. South Africa has produced four map versions, from a rudimentary map of 34 biozones informed by bathymetry data, to the latest version comprising 163 ecosystem types informed by 83 environmental and biodiversity datasets that aligns with the IUCN Global Ecosystem Typology. Data were unlocked through academic and industry collaborations; multi-disciplinary, multi-realm and multi-generational networks of practitioners; and targeted research to address key gaps. To advance toward a more transparent, reproducible and data-driven approach, limitations, barriers and opportunities for improvement were identified. Challenges included limited human and data infrastructure capacity to collate, curate and assimilate many data sources, covering a variety of ecosystem components, methods and scales. Five key lessons that are of relevance for others working to advance ecosystem classification and mapping, were distilled. These include (1) the benefits of iterative improvement; (2) the value of fostering relationships among a co-ordinated network of practitioners including early-career researchers; (3) strategically prioritizing and leveraging resources to build and curate key foundational biodiversity datasets and understand drivers of biodiversity pattern; (4) the need for developing, transferring and applying capacity and tools that enhance data quality, analytical workflows and outputs; and (5) the application of new technology and emerging statistical tools to improve the classification and prediction of biodiversity pattern. South Africa’s map of marine ecosystem types has been successfully applied in spatial biodiversity assessment, prioritization to support protected area expansion and marine spatial planning. These successes demonstrate the value of a co-ordinated network of practitioners who continually build an evidence base and iteratively improve ecosystem mapping while simultaneously growing ecological knowledge and informing changing priorities and policy.

IBSA_Net: A Network for Tomato Leaf Disease Identification Based on Transfer Learning with Small Samples

Tomatoes are a crop of significant economic importance, and disease during growth poses a substantial threat to yield and quality. In this paper, we propose IBSA_Net, a tomato leaf disease recognition network that employs transfer learning and small sample data, while introducing the Shuffle Attention mechanism to enhance feature representation. The model is optimized by employing the IBMax module to increase the receptive field and adding the HardSwish function to the ConvBN layer to improve stability and speed. To address the challenge of poor generalization of models trained on public datasets to real environment datasets, we developed an improved PlantDoc++ dataset and utilized transfer learning to pre-train the model on PDDA and PlantVillage datasets. The results indicate that after pre-training on the PDDA dataset, IBSA_Net achieved a test accuracy of 0.946 on a real environment dataset, with an average precision, recall, and F1-score of 0.942, 0.944, and 0.943, respectively. Additionally, the effectiveness of IBSA_Net in other crops is verified. This study provides a dependable and effective method for recognizing tomato leaf diseases in real agricultural production environments, with the potential for application in other crops.