Predictive Mapping Research Articles

Application of Deep Learning for Reservoir Porosity Prediction and Self Organizing Map for Lithofacies Prediction

While there is a connection between petrophysical logs and reservoir porosity, finding analytical solutions for this relationship is still difficult. This paper presents a novel approach for forecasting porosity and lithofacies by using a convolutional neural network (CNN) model in conjunction with a bi-directional long short-term memory (BLSTM) network. The BLSTM network uses a self-organizing map (SOM) technique to form connections between input and destination data. The SOM is used to organize depth intervals with similar facies into four separate clusters, each exhibiting internal consistency in petrophysical parameters. The CNN is responsible for extracting spatial characteristics, while the BLSTM network gathers comprehensive spatiotemporal components, guaranteeing that the model accurately represents the spatiotemporal aspects of log data. The accuracy of the model was verified by analyzing simulation logging data. The findings indicate that the BLSTM network model successfully recovers significant characteristics from logging data, resulting in improved estimate accuracy. In addition, Facies-01 has lower gamma ray levels in comparison to other facies. Facies-01 is also suggestive of pristine sandstone formations, which are greatly sought as reservoir rocks. The BLSTM network model is effective in predicting physical characteristics of reservoirs, offering a new method for precise reservoir characterization parameter prediction.

Predicting rock art sites in the Pajeú watershed, Brazil

This paper presents an Archaeological Predictive Model (APM) to predict rock art archaeological sites in the Pajeú Watershed, a semiarid region in Pernambuco, Brazil. The model uses Machine Learning (ML) algorithms and re-sampling techniques to account for the unbalanced data set of rock art sites and test different inductive methods for predicting site location. The results show a satisfactory statistical evaluation, with high true positive rates with all ML algorithms and resampling techniques used, indicating a high potential for predicting rock art site locations. The predictive maps generated from the model output, show that certain features, such as aspect, elevation and the distance to different lithologies, are particularly important. The overall model's performance could be corroborated with a test in another semi-arid region, next to the Pajeú watershed, where areas with high favorability of finding rock art sites are predicted near to already known archaeological sites.

Predicting peritumoral glioblastoma infiltration and subsequent recurrence using deep-learning-based analysis of multi-parametric magnetic resonance imaging.

Glioblastoma (GBM) is the most common and aggressive primary adult brain tumor. The standard treatment approach is surgical resection to target the enhancing tumor mass, followed by adjuvant chemoradiotherapy. However, malignant cells often extend beyond the enhancing tumor boundaries and infiltrate the peritumoral edema. Traditional supervised machine learning techniques hold potential in predicting tumor infiltration extent but are hindered by the extensive resources needed to generate expertly delineated regions of interest (ROIs) for training models on tissue most and least likely to be infiltrated. We developed a method combining expert knowledge and training-based data augmentation to automatically generate numerous training examples, enhancing the accuracy of our model for predicting tumor infiltration through predictive maps. Such maps can be used for targeted supra-total surgical resection and other therapies that might benefit from intensive yet well-targeted treatment of infiltrated tissue. We apply our method to preoperative multi-parametric magnetic resonance imaging (mpMRI) scans from a subset of 229 patients of a multi-institutional consortium (Radiomics Signatures for Precision Diagnostics) and test the model on subsequent scans with pathology-proven recurrence. Leave-one-site-out cross-validation was used to train and evaluate the tumor infiltration prediction model using initial pre-surgical scans, comparing the generated prediction maps with follow-up mpMRI scans confirming recurrence through post-resection tissue analysis. Performance was measured by voxel-wised odds ratios (ORs) across six institutions: University of Pennsylvania (OR: 9.97), Ohio State University (OR: 14.03), Case Western Reserve University (OR: 8.13), New York University (OR: 16.43), Thomas Jefferson University (OR: 8.22), and Rio Hortega (OR: 19.48). The proposed model demonstrates that mpMRI analysis using deep learning can predict infiltration in the peri-tumoral brain region for GBM patients without needing to train a model using expert ROI drawings. Results for each institution demonstrate the model's generalizability and reproducibility.

Novel dimensionless predictive flow pattern map for HFOs inside microscale enhanced tubes

Models designated to predict flow patterns in microscale geometries with enhanced surfaces such as micro-finned tubes are scarce in the literature and new low-GWP HFOs could benefit from the utilization of such geometries. Therefore, HFO1234ze(E)’s condensation flow patterns were subjected to visualization inside micro-finned tubes ranging from 4 to 7 mm outer diameters with various geometrical figures. The saturation temperature was set to 30 °C, vapor qualities ranged in the scope of 0.01 to 0.9, and mass fluxes in the magnitude of 100 to 400 kg m-2 s-1. Four distinctive flow patterns are observed, namely intermittent, annular, wavy-stratified, and transitional. Stratification only transpired at low mass fluxes (mainly below 100 kg m-2 s-1) and intermittent flow was only present at vapor qualities close to full condensation. The range in which transitional flow is observable shrinks with a progressive trend of mass flux. The impact of diameter was observed to be negligible, however, a more meticulous assessment highlights smaller ranges of vapor qualities in which transitional flow is present for the tube of 5 mm OD whose helix angle is substantially larger. Datapoints were juxtaposed to previous models of Doretti et al., Chen et al., Mandhane et al., and Jige et al. and the results attested to an inadequacy of accurate predictions made for tube of 4 mm OD. Noting the absence of surface tension force in the aforementioned maps, a novel flow pattern map based on modified Froude versus modified Weber numbers provided an accurate prediction for the three cases under study. Ultimately the model was also deemed fairly suitable for visualization datasets collected from literature.

Spatial Mapping of Soil CO2 Flux in the Yellow River Delta Farmland of China Using Multi-Source OpticalRemote Sensing Data

The spatial prediction of soil CO2 flux is of great significance for assessing regional climate change and high-quality agricultural development. Using a single satellite to predict soil CO2 flux is limited by climatic conditions and land cover, resulting in low prediction accuracy. To this end, this study proposed a strategy of multi-source spectral satellite coordination and selected seven optical satellite remote sensing data sources (i.e., GF1-WFV, GF6-WFV, GF4-PMI, CB04-MUX, HJ2A-CCD, Sentinel 2-L2A, and Landsat 8-OLI) to extract auxiliary variables (i.e., vegetation indices and soil texture features). We developed a tree-structured Parzen estimator (TPE)-optimized extreme gradient boosting (XGBoost) model for the prediction and spatial mapping of soil CO2 flux. SHapley additive explanation (SHAP) was used to analyze the driving effects of auxiliary variables on soil CO2 flux. A scatter matrix correlation analysis showed that the distributions of auxiliary variables and soil CO2 flux were skewed, and the linear correlations between them (r < 0.2) were generally weak. Compared with single-satellite variables, the TPE-XGBoost model based on multiple-satellite variables significantly improved the prediction accuracy (RMSE = 3.23 kg C ha−1 d−1, R2 = 0.73), showing a stronger fitting ability for the spatial variability of soil CO2 flux. The spatial mapping results of soil CO2 flux based on the TPE-XGBoost model revealed that the high-flux areas were mainly concentrated in eastern and northern farmlands. The SHAP analysis revealed that PC2 and the TCARI of Sentinel 2-L2A and the TVI of HJ2A-CCD had significant positive driving effects on the prediction accuracy of soil CO2 flux. The above results indicate that the integration of multiple-satellite data can enhance the reliability and accuracy of spatial predictions of soil CO2 flux, thereby supporting regional agricultural sustainable development and climate change response strategies.

Local Field-Scale Winter Wheat Yield Prediction Using VENµS Satellite Imagery and Machine Learning Techniques

Reliable and accurate crop yield prediction at the field scale is critical for meeting the global demand for reliable food sources. In this study, we tested the viability of VENμS satellite data as an alternative to other popular and publicly available multispectral satellite data to predict winter wheat yield and produce a yield prediction map for a field located in southwestern Ontario, Canada, in 2020. Random forest (RF) and support vector regression (SVR) were the two machine learning techniques employed. Our results indicate that machine learning models paired with vegetation indices (VIs) derived from VENμS imagery can accurately predict winter wheat yield 1~2 months prior to harvest, with the most accurate predictions achieved during the early fruit development stage. While both machine learning approaches were viable, SVR produced the most accurate prediction with an R2 of 0.86 and an RMSE of 0.3925 t/ha using data collected from tillering to the early fruit development stage. NDRE-1, NDRE-2, and REP from various growth stages were ranked among the top seven variables in terms of importance for the prediction. These findings provide valuable insights into using high-resolution satellites as tools for non-destructive yield potential analysis.

Slaughterhouse Wastewater Properties Assessment by Modern and Classic Methods

Advanced 1H Nuclear Magnetic Resonance (NMR) relaxometry and diffusometry methods and VIS-nearIR spectroscopy combined with pH, electrical conductivity (EC) and totally dissolved solids (TDSSs) measurements were used to assess the properties of wastewater collected from a chicken slaughterhouse in each step of the treatment process (wastewater before treatment, biologically treated wastewater, chemically treated wastewater and discharged wastewater) and from sludge. The 1H NMR Carr–Purcell–Meiboom–Gill (CPMG) and Pulsed-Gradient-Stimulated-Echo (PGSE) decay curves recorded for all samples of wastewater were analyzed by inverse Laplace transform (ILT) to obtain the distributions of transverse relaxation times T2 and diffusion coefficient D. The VIS-nearIR total absorbance, T2-values, D-values, pH, EC and TDSS parameters were used for statistical analysis in principal component (PCA). The 1H T2-distributions measured for the slaughterhouse wastewater lie in two main regions reflecting the number of dissolved solids or the distribution of undissolved solids. The PCA analysis successfully differentiates between polluted and less polluted wastewaters and sludge. The wastewater treatment applied by the slaughterhouse is efficient. The recommended methods for wastewater monitoring are the NMR T2- and D-distributions and EC, TDSSs and NMR-D diffusion coefficient. Finally, Machine Learning algorithms are used to provide prediction maps of wastewater treatment stage.

A robust ensemble of hybrid and bivariate statistical models for flood prediction mapping in Lower Damodar River Basin of India

This research explores flood prediction in the Lower Damodar River Basin (LDRB) using a hybrid ensemble of a Naïve Bayes Tree (NBT) and five bivariate statistical models such as Evidential Belief Function (EBF), Index of Entropy (IOE), Frequency Ratio (FR), Statistical Index (SI), and Modified Information Value (MIV). A total of 348 flood locations and 15 conditioning factors including hydrological, topographical and land cover were considered for this analysis. To ensure the precision of model predictions, a multicollinearity assessment was executed. Receiver operating characteristic (ROC) curve, area under curve (AUC), mean absolute error (MAE), mean squared error (MSE), and root mean squared error (RMSE) were performed to compare and asses each of the models all. The results reveal that all models performed well in creating flood hazard maps with AUC >0.8 and RMSE <0.4. FR and EBF models demonstrated the highest predictive accuracy (AUC=0.85), followed by the IOE, SI, and MIV models. The ensemble of NBT with bivariate models shows promising results, showcasing reduced error metrics and improved accuracy for the IOE, SI, and MIV models. This study highlights the potential of ensemble models in flood hazard prediction, offering valuable insights for global flood risk management. The successful application of these data-driven models showcases their importance in forecasting flood risks, aiding decision-makers and planners in developing more effective flood mitigation strategies.

Predictive grid coding in the medial entorhinal cortex.

The entorhinal cortex represents allocentric spatial geometry and egocentric speed and heading information required for spatial navigation. However, it remains unclear whether it contributes to the prediction of an animal's future location. We discovered grid cells in the medial entorhinal cortex (MEC) that have grid fields representing future locations during goal-directed behavior. These predictive grid cells represented prospective spatial information by shifting their grid fields against the direction of travel. Predictive grid cells discharged at the trough phases of the hippocampal CA1 theta oscillation and, together with other types of grid cells, organized sequences of the trajectory from the current to future positions across each theta cycle. Our results suggest that the MEC provides a predictive map that supports forward planning in spatial navigation.

Risk profiling of tobacco epidemic and estimated number of smokers living in China: a cross-sectional study based on PBICR

BackgroundEvidence on the prevalence of smoking in China remains insufficient, with most previous studies focusing on a single region. However, smoking prevalence exhibits significant inequalities across the entire country. This study aimed to evaluate the risk of tobacco prevalence across the country, taking into account spatial inequalities.MethodsThe data used in this study were collected in 23 provinces, 5 autonomous regions, and 4 municipalities directly under the central government in 2022. Large population survey data were used, and a Bayesian geostatistical model was employed to investigate smoking prevalence rates across multiple spatial domains.FindingsSignificant spatial variations were observed in smokers and exposure to secondhand smoke across China. Higher levels of smokers and secondhand smoke exposure were observed in western and northeastern regions. Additionally, the autonomous region of Tibet, Shanghai municipality, and Yunnan province had the highest prevalence of smokers, while Tibet, Qinghai province, and Yunnan province had the highest prevalence of exposure to secondhand smoke.ConclusionWe have developed a model-based, high-resolution nationwide assessment of smoking risks and employed rigorous Bayesian geostatistical models to help visualize smoking prevalence predictions. These prediction maps provide estimates of the geographical distribution of smoking, which will serve as strong evidence for the formulation and implementation of smoking cessation policies.HighlightsOur study investigated the prevalence of smokers and exposure to secondhand smoke in different spatial areas of China and explored various factors influencing the smoking prevalence. For the first time, our study applied Bayesian geostatistical modeling to generate a risk prediction map of smoking prevalence, which provides a more intuitive and clear understanding of the spatial disparities in smoking prevalence across different geographical regions, economic levels, and development status. We found significant spatial variations in smokers and secondhand smoke exposure in China, with higher rates in the western and northeastern regions.

Tackling global biogeography and drivers of soil microbial dehalogenation traits and taxa: Insights from metagenomic profiling based on a curated dehalogenase database

To identify microbial resources for dehalogenation, develop effective remediation strategies, and reveal their significance in “One Health”, it is crucial to understand the occurrence, distribution, and drivers of soil dehalogenation functional traits and taxonomy groups at a broad scale, which is currently not well understood. To address the gaps, we characterized the biogeography of both dehalogenation traits and taxa assigned to six dehalogenation pathways, by metagenomic profiling global 4821 soils from eight habitats, based on a manually curated dehalogenase database (DhgaseDB). We found dehalogenation genes and microbes assigned to different pathways are everywhere, but varied consistently across habitats. The similarity of dehalogention traits and taxa composition declines with geographic distance, and that patterns are strongly correlated with geo-environmental factors. We identified anthropogenic organohalide pesticide inputs as the most influential factor on dehalogenation gene abundance, while soil properties, particularly pH, exert a larger impact on dehalogenation taxa diversity. Ultimately, we generated predictive maps of soil dehalogenation gene abundance and taxa diversity for the first time, highlighting the microbial dehalogenation hotpots in East Asia, Australia, Southern Africa, and coastal regions. Collectively, our study highlights the significant role of various microbial dehalogenation processes in organohalide biotransformation and environmental microecology, providing the necessary methodological basis for a deeper comprehension of the underlying mechanisms, thereby contributing to the advancement of tailored strategies for organohalide remediation.

A data-driven approach to identifying PFAS water sampling priorities in Colorado, United States.

Per and polyfluoroalkyl substances (PFAS), a class of environmentally and biologically persistent chemicals, have been used across many industries since the middle of the 20th century. Some PFAS have been linked to adverse health effects. Our objective was to incorporate known and potential PFAS sources, physical characteristics of the environment, and existing PFAS water sampling results into a PFAS risk prediction map that may be used to develop a PFAS water sampling prioritization plan for the Colorado Department of Public Health and Environment (CDPHE). We used random forest classification to develop a predictive surface of potential groundwater contamination from two PFAS, perfluorooctane sulfonate (PFOS) and perfluorooctanoate (PFOA). The model predicted PFAS risk at locations without sampling data into one of three risk categories after being "trained" with existing PFAS water sampling data. We used prediction results, variable importance ranking, and population characteristics to develop recommendations for sampling prioritization. Sensitivity and precision ranged from 58% to 90% in the final models, depending on the risk category. The model and prioritization approach identified private wells in specific census blocks, as well as schools, mobile home parks, and public water systems that rely on groundwater as priority sampling locations. We also identified data gaps including areas of the state with limited sampling and potential source types that need further investigation. This work uses random forest classification to predict the risk of groundwater contamination from two per- and polyfluoroalkyl substances (PFAS) across the state of Colorado, United States. We developed the prediction model using data on known and potential PFAS sources and physical characteristics of the environment, and "trained" the model using existing PFAS water sampling results. This data-driven approach identifies opportunities for PFAS water sampling prioritization as well as information gaps that, if filled, could improve model predictions. This work provides decision-makers information to effectively use limited resources towards protection of populations most susceptible to the impacts of PFAS exposure.

Predictive mapping and analysis of aging severity for acrylic resin coatings in global marine environments

Predictive mapping and analysis of aging severity for acrylic resin coatings in global marine environments

A novel dynamic predictive control model for variable air volume air conditioning systems using deep learning

Model-based predictive control has proven effective for managing indoor temperature and humidity in variable air volume (VAV) air conditioning systems. However, delays in temperature and humidity adjustments in response to changes in internal and external environmental conditions can impede high-performance operation. This study introduces a dynamic predictive control model for the multi-zone indoor temperature and humidity of VAV systems, designed to predict and control these parameters in real-time. Utilizing the RC network two-node wall structure method and backward finite difference scheme, a multi-zone building model is developed. Coupled with a deep fuzzy cognitive map (DFCM) for temperature and humidity prediction, and controlled by a PID controller, the model dynamically regulates the opening degrees of terminal air dampers and chilled water valves. Implemented on a networked control platform in a large commercial building, the model consistently maintained indoor temperature and humidity within ±0.5 °C and ±0.1 g/(kg dry air) of the target values under varying conditions, demonstrating substantial improvements in stability and operational efficiency. The research enhances the predictability and control of HVAC systems through advanced algorithmic integration, improving real-time indoor climate management in complex building environments. This method offers a practical improvement in HVAC control technologies, which could be beneficial for applications in commercial building management systems, providing a useful tool for supporting energy efficiency and sustainability in modern infrastructures.

Evaluation of accurate measurement methods for radon exhalation rate with advection effect and application in field

In decommissioning of a uranium tailings pond, radon exhalation rates on a beach surface should meet regulatory standards. Accurate measurements of the radon exhalation rate are demanded. However, current studies fail to consider the impact of advection under temperature variations or pressure gradients caused by gas movement on measurements using an accumulation chamber. Two proposed methods were therefore evaluated to accurately measure radon exhalation rates on the loose medium surface under advective conditions. Repeated experiments were conducted on a laboratory experimental platform filled with uranium tailings sand under advective flow rates of 0.03 and 0.3 L/min to validate the stability and reliability. Deviations between measured and true values were 0.1–6.1 % and 6.3–29.2 % for the two methods, respectively. Subsequently, numerical simulation was used to analyze defects of traditional methods and mechanisms of the new methods. In a field study, all methods were compared, and a predictive map of radon exhalation rates was created using interpolated data from 20 random sites using the new method. Results from the proposed methods, compared with traditional ones, were closer to true values under advective conditions, and accurate assessment of beach surface treatment was expected.

Derivation of Predictive Layers Using Regional Till Geochemistry Data for Mineral Potential Mapping of the REE Line of Bergslagen, Central Sweden

With the increasing need for rare-earth elements (REEs) to reach the goals of the ongoing green energy transition, new and innovative methods are needed to identify new primary resources of these critical metals. This study explores the potential to use a non-biased, uniform till dataset to generate evidentiary layers that describe these critical factors and geochemical anomalies to aid mineral potential mapping (MPM) for REEs using machine-assisted methods. The till samples used in this study were collected from the “REE Line”, a sub-region within the Bergslagen lithotectonic province, Sweden, where numerous REE mineralizations occur. Multiple approaches were used in this study to isolate geochemical anomalies using multivariate methods, namely principal component analysis (PCA) and K-means clustering. Additional factors for classifying till samples were also tested, including alteration indices. Using known REE occurrences in Bergslagen as validation points, the results demonstrated the usefulness of multivariate methods applied to till geochemistry for predictive bedrock mapping, and to identify potential areas of REE mineralization within the REE line. The results of the alteration indices showed that the till geochemistry demonstrated similar levels of alteration when compared to the underlying bedrock, allowing for a regional alteration map to be generated. These results show that regional-scale till sampling can provide low-cost data for mineral exploration at the regional scale and generate usable evidentiary layers for GIS-based MPM.

Machine learning applied to species occurrence and interactions: the missing link in biodiversity assessment and modelling of Antarctic plankton distribution

BackgroundPlankton is the essential ecological category that occupies the lower levels of aquatic trophic networks, representing a good indicator of environmental change. However, most studies deal with distribution of single species or taxa and do not take into account the complex of biological interactions of the real world that rule the ecological processes.ResultsThis study focused on analyzing Antarctic marine phytoplankton, mesozooplankton, and microzooplankton, examining their biological interactions and co-existences. Field data yielded 1053 biological interaction values, 762 coexistence values, and 15 zero values. Six phytoplankton assemblages and six copepod species were selected based on their abundance and ecological roles. Using 23 environmental descriptors, we modelled the distribution of taxa to accurately represent their occurrences. Sampling was conducted during the 2016–2017 Italian National Antarctic Programme (PNRA) ‘P-ROSE’ project in the East Ross Sea. Machine learning techniques were applied to the occurrence data to generate 48 predictive species distribution maps (SDMs), producing 3D maps for the entire Ross Sea area. These models quantitatively predicted the occurrences of each copepod and phytoplankton assemblage, providing crucial insights into potential variations in biotic and trophic interactions, with significant implications for the management and conservation of Antarctic marine resources. The Receiver Operating Characteristic (ROC) results indicated the highest model efficiency, for Cyanophyta (74%) among phytoplankton assemblages and Paralabidocera antarctica (83%) among copepod communities. The SDMs revealed distinct spatial heterogeneity in the Ross Sea area, with an average Relative Index of Occurrence values of 0.28 (min: 0; max: 0.65) for phytoplankton assemblages and 0.39 (min: 0; max: 0.71) for copepods.ConclusionThe results of this study are essential for a science-based management for one of the world’s most pristine ecosystems and addressing potential climate-induced alterations in species interactions. Our study emphasizes the importance of considering biological interactions in planktonic studies, employing open access and machine learning for measurable and repeatable distribution modelling, and providing crucial ecological insights for informed conservation strategies in the face of environmental change.

Predicting forest fire probability in Similipal Biosphere Reserve (India) using Sentinel-2 MSI data and machine learning

The global escalation in forest fires, characterized by increasing frequency and severity, results from a complex interplay of natural and anthropogenic factors, exacerbated by climate change. These fires devastate habitats, threaten species, reduce biodiversity, disrupt natural cycles, and harm local ecosystems. The impacts are particularly damaging in biological reserves. The Similipal Biosphere Reserve (SBR) in Odisha State is one of India’s major forest fire hotspots, experiencing forest fires almost every year. The objective of this study is to develop a predictive model using Sentinel-2 MSI data and machine learning (ML) techniques to estimate the probability of forest fires in the SBR, India, thereby enhancing disaster management and prevention in the region. This research maps and quantifies forest fire intensity by leveraging ML algorithms, namely Extreme Gradient Boosting (XGBoost), Gradient Boosting Machine (GBM), Support Vector Machine (SVM), and Random Forest (RF). To develop a Forest Fire Probability (FFP) map, twenty conditioning factors, along with pre- and post-fire Normalized Burn Ratio (NBR) and delta Normalized Burn Ratio (dNBR), were utilized. Furthermore, four statistical methods—Mean Absolute Error, Mean Square Error, Root Mean Square Error, and Overall Accuracy—were employed to analyze the FFP. The results were validated using the Area Under Curve (AUC) method. The analysis identifies 2021 as the year with the highest incidence of forest fires, accounting for 29.19% of the occurrences. Among the models, the GBM exhibits superior performance, highlighting its efficacy in handling large, multidimensional datasets. Predictive mapping suggests that approximately 1400–1500 km2, or 25–30% of the studied area, faces a high to very high risk of forest fires.

Habitat suitability modelling to predict the distribution of deep coral ecosystems: The case of Linosa Island (southern Mediterranean Sea, Italy)

In areas with limited field data, predictive habitat mapping is a valuable method for elucidating species-environment relationships and enhancing our knowledge of the spatial distribution and complexity of benthic habitats. Species distribution models (SDMs) can be an important tool to support in science-based ecosystem management. The availability of direct observations of mesophotic species, including gorgonians and black corals, during costly surveys is generally limited. Therefore, predicting the distribution of mesophotic species in relation to key physical parameters of the seafloor would help improving conservation strategies in existing and new Marine Protected Areas (MPAs). This study aims to assess the distribution of gorgonians and black corals off Linosa Island, in the Strait of Sicily, a biogeographic boundary area between the western and eastern Mediterranean. The volcanic island of Linosa represents a small, naturally preserved area, with very limited human pressure, hosting rich marine benthic biodiversity on its wide submarine portions. Distribution of the most common coral species off Linosa Island was modelled under an SDM framework, relying on direct observations collected during two research cruises in 2016 and 2017 and a series of terrain parameters acquired through geophysical techniques. We used the so-called “ensemble of small models” approach to calibrate SDMs, which achieved fair-to-excellent results (AUC >0.7). In addition to identifying depth as the primary factor influencing coral distribution, our study also highlighted ruggedness as a significant terrain variable. Specifically, the depth range of 110–230 m emerged as the critical parameter determining habitat suitability for all modelled species, also highlighting peculiar and specie–specific habitat requirements.

Predictive mapping of surface roughness in turning of hardened AISI 4340 using carbide tools

This study presents a novel approach to predict surface roughness in the hard turning of AISI 4340 steel using carbide tools, aimed to develop a comprehensive predictive map. The hypothesis that surface roughness can be accurately predicted using a linear regression model was tested and confirmed. Experimental results showed surface roughness in the range of 1.946 to 5.636 microns. Statistical analysis revealed a normal distribution of surface roughness data with linear regression as the best-fit model, significantly determined by feed rate and explaining 98.41% of the variance. Machine learning validated this model, achieving high prediction accuracy (R² = 96.91%, MSE = 0.058, RMSE = 0.242). The innovative predictive map, created using a full factorial design, demonstrated a strong agreement between predicted and validated values. This work highlights the potential of integrating statistical and machine learning techniques for precise surface roughness prediction, recommending industrial validation to enhance machining productivity.