Sampling Grid Research Articles

Efficient Supervised Pretraining of Swin-Transformer for Virtual Staining of Microscopy Images.

Fluorescence staining is an important technique in life science for labeling cellular constituents. However, it also suffers from being time-consuming, having difficulty in simultaneous labeling, etc. Thus, virtual staining, which does not rely on chemical labeling, has been introduced. Recently, deep learning models such as transformers have been applied to virtual staining tasks. However, their performance relies on large-scale pretraining, hindering their development in the field. To reduce the reliance on large amounts of computation and data, we construct a Swin-transformer model and propose an efficient supervised pretraining method based on the masked autoencoder (MAE). Specifically, we adopt downsampling and grid sampling to mask 75% of pixels and reduce the number of tokens. The pretraining time of our method is only 1/16 compared with the original MAE. We also design a supervised proxy task to predict stained images with multiple styles instead of masked pixels. Additionally, most virtual staining approaches are based on private datasets and evaluated by different metrics, making a fair comparison difficult. Therefore, we develop a standard benchmark based on three public datasets and build a baseline for the convenience of future researchers. We conduct extensive experiments on three benchmark datasets, and the experimental results show the proposed method achieves the best performance both quantitatively and qualitatively. In addition, ablation studies are conducted, and experimental results illustrate the effectiveness of the proposed pretraining method. The benchmark and code are available at https://github.com/birkhoffkiki/CAS-Transformer.

Accuracy and sensitivity of soil erosion estimation using 137Cs technology: A statistical perspective

Random spatial variation of the 137Cs inventory is the principal contributor to uncertainty in soil erosion estimation using 137Cs technology. A statistically sound sampling design is imperative for obtaining reliable soil erosion estimations. The objectives of this study are to: 1) characterize the effect of sample size on the estimates of mean inventories; 2) evaluate the sensitivity of the estimated soil redistribution to sample size using a new 137Cs model; 3) compare erosion detection limits among five conversion models; and 4) assess three spatial interpolation methods. Two small watersheds were sampled in an irregular grid design, and 30 samples were taken on a reference site. A moving window scheme was used to compute the 7-point and 11-point means of 137Cs inventories. The Welch’s modified t-test was used to test the mean estimates and to compute the detection limits for the replicated samples. Five models were selected to convert 137Cs inventories to soil redistribution rates. Since most single samples without replication yielded insignificant test results, the 137Cs technique cannot be used to estimate soil erosion at a single point due to random spatial variation. With 30 reference samples, 7 replicate samples on the sampling site provided reliable estimations of 137Cs inventories and soil redistribution. The spatial patterns of the estimated 137Cs inventories and soil redistribution became more regular and systematic as sample size increased, which agreed increasingly well with topography and surface hydrology. Replicate samples can be taken from each landform element or slope position where the erosion rate is expected to be uniform. Alternatively, a systematic grid sampling scheme can be used, and the nearest neighbors can be treated as replicates to calculate moving averages. The Type I and II errors with replicated samples showed that all sampling positions should be included in calculating watershed or area mean erosion rates even though the Type I errors indicated insignificant redistribution rates for some samples. Due to the lack of spatial autocorrelation, spline and inverse distance weighting methods performed better than kriging for interpolation. Detection limits and magnitudes of soil redistribution varied substantially with conversion models, depending on their sensitivities to changes in 137Cs inventories.

Effect of Starch and Paperboard Reinforcing Structures on Insulative Fiber Foam Composites.

Single-use plastic foams are used extensively as interior packaging to insulate and protect items during shipment but have come under increasing scrutiny due to the volume sent to landfills and their negative impact on the environment. Insulative compression molded cellulose fiber foams could be a viable alternative, but they do not have the mechanical strength of plastic foams. To address this issue, a novel approach was used that combined the insulative properties of cellulose fiber foams, a binder (starch), and three different reinforcing paperboard elements (angular, cylindrical, and grid) to make low-density foam composites with excellent mechanical strength. Compression molded foams and composites had a consistent thickness and a smooth, flat finish. Respirometry tests showed the fiber foams mineralized in the range of 37 to 49% over a 46 d testing period. All of the samples had relatively low density (Dd) and thermal conductivity (TC). The Dd of samples ranged from 33.1 to 64.9 kg/m3, and TC ranged from 0.039 to 0.049 W/mk. The addition of starch to the fiber foam (FF+S) and composites not only increased Dd, drying time (Td), and TC by an average of 18%, 55%, and 5.5%, respectively, but also dramatically increased the mechanical strength. The FF+S foam and paperboard composites had 240% and 350% higher average flexural strength (σfM) and modulus (Ef), respectively, than the FF-S composites. The FF-S grid composite and all the FF+S foam and composite samples had equal or higher σfM than EPS foam. Additionally, FF+S foam and paperboard composites had 187% and 354% higher average compression strength (CS) and modulus (Ec), respectively, than the FF-S foam and composites. All the paperboard composites for both FF+S and FF-S samples had comparable or higher CS, but only the FF+S cylinder and grid samples had greater toughness (Ωc) than EPS foam. Fiber foams and foam composites are compatible with existing paper recycling streams and show promise as a biodegradable, insulative alternative to EPS foam internal packaging.

Arbitrary-Scale Point Cloud Upsampling by Voxel-Based Network with Latent Geometric-Consistent Learning

Recently, arbitrary-scale point cloud upsampling mechanism became increasingly popular due to its efficiency and convenience for practical applications. To achieve this, most previous approaches formulate it as a problem of surface approximation and employ point-based networks to learn surface representations. However, learning surfaces from sparse point clouds is more challenging, and thus they often suffer from the low-fidelity geometry approximation. To address it, we propose an arbitrary-scale Point cloud Upsampling framework using Voxel-based Network (PU-VoxelNet). Thanks to the completeness and regularity inherited from the voxel representation, voxel-based networks are capable of providing predefined grid space to approximate 3D surface, and an arbitrary number of points can be reconstructed according to the predicted density distribution within each grid cell. However, we investigate the inaccurate grid sampling caused by imprecise density predictions. To address this issue, a density-guided grid resampling method is developed to generate high-fidelity points while effectively avoiding sampling outliers. Further, to improve the fine-grained details, we present an auxiliary training supervision to enforce the latent geometric consistency among local surface patches. Extensive experiments indicate the proposed approach outperforms the state-of-the-art approaches not only in terms of fixed upsampling rates but also for arbitrary-scale upsampling. The code is available at https://github.com/hikvision-research/3DVision

Assessing the performance of wind erosion prediction models of RWEQ and SWEEP by direct measurements in an arid region

Wind erosion field studies allow researchers to conduct simulations under natural conditions, thus play a significant role in investigating wind erosion processes. Direct measurements are also used to validate and improve existing erosion models. There is a need to develop sustainable land-use practices and management strategies and to assess potential current and future trends of wind erosion risk in arid and semi-arid regions. This study assessed wind erosion-induced soil losses in 11 different wind cases encountered within the wheat-fallow production system in a semi-arid region of Türkiye between 2017 and 2020 to eliminate the wind erosion threats on soil resources. Direct measurements and two estimation models (RWEQ and SWEEP) were employed to assess soil loss. Grid sampling method and BEST® sediment traps were utilized to measure wind-blown soil in the field directly. Present findings revealed that wind erosion threatened both plots in different seasons. The wheat-cultivated plot was highly susceptible to wind erosion in autumn when winter cereals are sown, while the fallow plot showed susceptibility in spring. Throughout the entire measurement period (3-years), wind erosion-induced soil transportation was measured as 1383 kg ha−1 for fallow plots and 2548 kg ha−1 for wheat-cultivated plots. Model performance was assessed through correlations (r), root mean square error (RMSE) and d-test results. The results demonstrated that the calibrated RWEQ (RWEQcal) model provided reasonably accurate estimates of wind erosion for both fallow and wheat-cultivated lands, with relatively low RMSE values of 0.34 and 0.46, d-test values of 0.26 and 0.27, and correlation coefficients of 0.69 and 0.85, respectively. Model results highlighted the importance of model calibration for local conditions; otherwise, they may produce underestimated and overestimated results.

A novel bioprospecting strategy via 13C-based high-throughput probing of active methylotrophs inhabiting oil reservoir surface soil

Methane-oxidizing bacteria (MOB) have long been considered as a microbial indicator for oil and gas prospecting. However, due to the phylogenetically narrow breath of ecophysiologically distinct MOB, classic culture-dependent approaches could not discriminate MOB population at fine resolution, and accurately reflect the abundance of active MOB in the soil above oil and gas reservoirs. Here, we presented a novel microbial anomaly detection (MAD) strategy to quantitatively identify specific indicator methylotrophs in the surface soils for bioprospecting oil and gas reservoirs by using a combination of 13C-DNA stable isotope probing (SIP), high-throughput sequencing (HTS), quantitative PCR (qPCR) and geostatistical analysis. The Chunguang oilfield of the Junggar Basin was selected as a model system in western China, and type I methanotrophic Methylobacter was most active in the topsoil above the productive oil wells, while type II methanotrophic Methylosinus predominated in the dry well soils, exhibiting clear differences between non- and oil reservoir soils. Similar results were observed by quantification of Methylobacter pmoA genes as a specific bioindicator for the prediction of unknown reservoirs by grid sampling. A microbial anomaly distribution map based on geostatistical analysis further showed that the anomalous zones were highly consistent with petroleum, geological and seismic data, and validated by subsequent drilling. Over seven years, a total of 24 wells have been designed and drilled into the targeted anomaly, and the success rate via the MAD prospecting strategy was 83 %. Our results suggested that molecular techniques are powerful tools for oil and gas prospecting. This study indicates that the exploration efficiency could be significantly improved by integrating multi-disciplinary information in geophysics and geomicrobiology while reducing the drilling risk to a greater extent.

Defining management zones for a papaya culture considering soil chemical attributes by fuzzy c-means

In precision agriculture, one of the major problems is the need for a dense sampling grid. A solution to this problem has been the generation of management zones and the correlation between them. This work aims at defining management zones for a papaya culture using the fuzzy C-means grouping method, based on the following soil attributes: phosphorus, calcium, magnesium and potassium; and base saturation. The work was conducted in an approximately 1.2 ha area planted with a commercial culture of Carica papaya L. cultivated in typical Argisol found on the coastal terraces, in the municipality of São Mateus, ES, Brazil. The management zones were defined by the proposed method, based on the following analyses: soil chemical parameters P, Ca, Mg, and base saturation (V), analyzed together; soil chemical parameters P, Ca, Mg, and V, analyzed separately; and, the following combinations of soil chemical attributes Ca-Mg, Ca-V, Ca-Mg-V, P-Ca, P-Mg, P-V, P-Ca-Mg, P-Ca-V, Mg-V, and P-Ca-Mg-V. The generated management zones facilitated an understanding of the spatial variability of the attributes in the study area and the observed similarities between management zones generated from different attributes. In addition, the data were influenced especially by the P and V% attributes. Keywords: Carica papaya L, geostatistics, grouping analysis, precision agriculture.

Monitoring of North Island fantail / pīwakawaka fantail (Rhipidura fuliginosa placabilis Bangs, 1921, Rhipiduridae) distribution on Tiritiri Matangi Island: using several spatial methods for processing volunteer data

The Aotearoa / New Zealand fantail / pīwakawaka (Rhipidura fuliginosa Sparrman, 1787, Rhipiduridae) is an iconic species and conspicuous in a range of habitats. However, island populations of the species are said to fluctuate dramatically. This project set out to investigate the population dynamics of the Te ika a Maui / North Island subspecies Rhipidura fuliginosa placabilis Bangs, 1921) on Tiritiri Matangi Island using a volunteer-based monitoring scheme. To achieve this goal we developed a pilot sampling grid and determined spatial distribution characteristics using several approaches including geostatistical tools in ArcMap, species distribution modelling and occupancy modelling. Fantail presence data was recorded twice a year by volunteers assigned to specific grids at different seasons for two years on Tiritiri Matangi Island. Recorded count data show distinctive differences between the sampling periods and different areas of the island. Significant hotspots as well as clustering of abundance show different patterns with significantly higher abundance and widespread distribution during May compared to November. Spatial analysis identified vegetation, particularly replanted areas, as influencing the fantail counts. Results of species distribution modelling showed areas of the island suitable for the species and occupancy models further described the seasonal spatial characteristics of fantail. The effort also highlights the importance of volunteers in providing bird count data to generate the knowledge base required for the management of an island sanctuary.

Mapping of volcanic soil chemical properties with digital soil mapping after the prolonged eruption of Mt. Sinabung (2013-2020)

Volcanic soils in Indonesia cover an area of about 5.4 million ha and around 2.7 million ha in Sumatra. The prolonged eruptions of Mt. Sinabung in North Sumatra from 2013 to today eject pyroclastic materials, which blanketed and altered the soil surface. As a result, volcanic soil formed from these pyroclastic deposits. This study aims to map the chemical properties of volcanic soils after the prolonged eruption of Mt. Sinabung from 2013-2020. Thirty-four soil samples were collected at a depth of 0-20 cm according to the grid sampling system with an interval of 1x1 km covering an area of about 4,500 ha. Regression kriging (RK) was applied to predict the soil’s chemical properties and distribution spatially. The soil samples were air dried, sieved, and analyzed to determine soil pH (H2O and KCl), available, potential and retention P, organic carbon, total nitrogen, cation exchange capacity (CEC), and exchangeable basic cations. Soil pH (H2O) ranges from very acidic to neutral (4.14-6.52) and very acidic to acidic pH (KCl) (3.89-5.26), low to very high available-P (3.46-382.01 ppm), potential-P very low to very high (3.76-230.26 mg 100g-1), P-retention is categorized as very high (90-99%). The organic Carbon range from low to very high (1.73-13.05%), low to high total nitrogen (0.13-0.60%), low to high cation exchange capacity (11.78-97.71 cmolc kg-1), and exchangeable base cations are categorized as high with K values (1.60-2.98 cmolc kg-1). Na (3.72-7.45 cmolc kg-1) and Mg (5.79-12.15 cmolc kg-1) were categorized as high and Ca very low (0.039-0.157 cmolc kg-1). The estimated area of soils with pH between 4 and 5 is about 424.85 ha, and soils with pH between 5 and 6 covered an area of approximately 3,964.81 ha. Our findings suggest that persistent eruptions of Mt. Sinabung supply new plant nutrition, which can enrich and increase soil fertility in the future.

Analysis of Groundwater Salinity Levels in the Lampulo Coastal Area Kuta Alam Sub-District

The problem of seawater seepage in coastal areas can result in a decrease in groundwater quality due to the mixing of seawater and groundwater, so seawater seepage has an impact on increasing salinity, including in the research area. The research location is in the Lampulo beach area, Kuta Alam sub-district. Phreatic well water is generally not liked by residents of this coastal area because it tastes brackish. The high salinity of phreatic groundwater is caused by seawater seepage into alluvial land. The aim of this research is to analyze the level of groundwater salinity in the Lampulo coastal area, Kuta Alam sub-district. This research uses a grid sampling method and criteria for assessing groundwater salinity (o/oo) based on salt content. Groundwater salinity was measured randomly in 50 phreatic wells using a refractometer and marked with sample coordinates using the Avenza Maps application. Field measurement results show that there are variations in salinity values ranging from 0 to 32 ppt. Phreatic well water at the research location based on its salt content is grouped into three groups, namely fresh water, brackish water and salt water. For the freshwater group, a percentage value was obtained with a prevalence of 68%; for the brackish water group, a percentage value was obtained with a prevalence of 26%; and for the salt water group, a percentage value was obtained with a prevalence of 6%.Keywords: Salinity, Groundwater, Phreatic Well, Seawater Seepage, Grid Sampling

Efficient prediction of individual head-related transfer functions based on 3D meshes

Individual head-related transfer functions (HRTFs) are critical for binaural spatial audio rendering. In contrast to anthropometric parameters and pinnae images, 3D meshes allow for a more direct and comprehensive representation of the anthropometric structure, which provides highly effective inputs for modeling individualized HRTFs. This paper presents a neural network-based method for predicting individualized HRTFs in full space based on 3D meshes. Unlike many previous methods that estimate HRTF spectra at sampling grids or frequencies separately, the proposed model predicts the HRTF spectra of each vertical plane by considering the spectral correlation and continuity across adjacent sampling grids and frequencies. Evaluation results indicate that the proposed method enhances the prominence of peaks and notches in the obtained HRTF spectra and improves the speed and accuracy of HRTF individualization. The log spectral distortion of the proposed method is lower than that of state-of-the-art methods using anthropometric parameters and pinnae images. Further evaluation confirms that the proposed method requires significantly fewer points in 3D meshes when compared to numerical simulation methods. The evaluation based on localization models demonstrates that the HRTFs predicted by the proposed method are perceptually similar to the measured HRTFs.

Physically Informed Deep Learning Technique for Estimating Blood Flow Parameters in Four-Vessel Junction after the Fontan Procedure

This study introduces an innovative approach leveraging physics-informed neural networks (PINNs) for the efficient computation of blood flows at the boundaries of a four-vessel junction formed by a Fontan procedure. The methodology incorporates a 3D mesh generation technique based on the parameterization of the junction’s geometry, coupled with an advanced physically regularized neural network architecture. Synthetic datasets are generated through stationary 3D Navier–Stokes simulations within immobile boundaries, offering a precise alternative to resource-intensive computations. A comparative analysis of standard grid sampling and Latin hypercube sampling data generation methods is conducted, resulting in datasets comprising 1.1×104 and 5×103 samples, respectively. The following two families of feed-forward neural networks (FFNNs) are then compared: the conventional “black-box” approach using mean squared error (MSE) and a physically informed FFNN employing a physically regularized loss function (PRLF), incorporating mass conservation law. The study demonstrates that combining PRLF with Latin hypercube sampling enables the rapid minimization of relative error (RE) when using a smaller dataset, achieving a relative error value of 6% on the test set. This approach offers a viable alternative to resource-intensive simulations, showcasing potential applications in patient-specific 1D network models of hemodynamics.

Equalizer Parameters’ Adjustment Based on an Oversampled Channel Model for OFDM Modulation Systems

A physical model of a wireless transmission channel in the time domain usually consists of the main propagation path and only a few reflections. The reasonable assumptions made about the channel model can improve its parameters’ estimation by a greedy OFDM (Orthogonal Frequency Division Multiplexing) equalizer. The equalizer works flawlessly if delays between propagation paths are in the sampling grid. Otherwise, the channel impulse response loses its compressible characteristic and the number of coefficients to find increases. It is possible to get back to the simple channel model by data oversampling. The paper describes how the above idea helps the OMP (Orthogonal Matching Pursuit) algorithm estimate channel coefficients. The authors analyze the oversampling algorithm on the one hand to assess the influence of filtering function and signal resolution on the quality of the channel impulse response reconstruction. On the other hand, the abilities of the OMP algorithm are analyzed to distinguish components of the oversampled signal. Based on these analyses, we proposed modifications to the compressible channel’s impulse response reconstruction algorithm to minimize the number of transmission errors. A distinction was made between the filters used in the OMP search and channel reconstruction stages before calculating equalizer coefficients. Additionally, the results of the search stage were considered as elements within the groups.

A new approach for spatializing the Canadian National Forest Inventory (SCANFI) using Landsat dense time series

Accurate and fine-scale forest data are essential to improve natural resource management, particularly in the face of climate change. Here, we present SCANFI, the Spatialized CAnadian National Forest Inventory, which provides coherent, 30 m resolution 2020 wall-to-wall maps of forest attributes (land cover type, canopy height, crown closure, aboveground tree biomass, and main species composition). These maps were developed using the NFI photo-plot dataset, a systematic regular sample grid of photo-interpreted high-resolution imagery covering all of Canada's non-arctic landmass. SCANFI was produced using temporally harmonized summer and winter Landsat spectral imagery along with hundreds of tile-level regional models based on a multiresponse k-nearest neighbours and random forest imputation method. This tile-level approach revealed the importance of radiometric variables in predicting vegetation attributes, namely winter radiometry, as the large-scale climate gradients were controlled at the tile-level. SCANFI was validated with rigorous cross-validation analyses, which revealed robust model performance for structural attributes (biomass R2 = 0.76; crown closure R2 = 0.82; height R2 = 0.78) and tree species cover (e.g., Douglas fir R2 = 0.60). SCANFI attributes were also validated with several independent external products, ranging from ground plot-based tree species cover (e.g., black spruce R2 = 0.53) to satellite LiDAR height data products (e.g., crown closure R2 = 0.71). SCANFI total aboveground biomass trends also followed those published by other studies. The methodology presented herein can be used to map time series of these attributes, identify the original training points used to make any given prediction, as well as map additional variables associated with the NFI photo-plots that are challenging to map using traditional remote sensing approaches.

Confidence partitioning sampling filtering

AbstractThe confidence partitioning sampling filter (CPSF) method proposed in this paper is a novel approach for solving the generic nonlinear filtering problem. First, the confidence probability space (CPS) is defined, which restricts the state transition in a bounded and closed state space in the recursive Bayesian filtering. In the posterior CPS, the weighted grid samples, represented the posterior PDF, are obtained by using the partitioning sampling technique (PST). Each weighted grid sample is treated as an impulse function. The approximate expression of the posterior PDF, as key for the PST implementation, is obtained by using the properties of the impulse function in the integral operation. By executing the selection of the CPS and the PST step repeatedly, the CPSF framework is formed to achieve the approximation of the recursive Bayesian filtering. Second, the difficulty of the CPSF framework implementation lies in obtaining the real posterior CPS. Therefore, the space intersection (SI) method is suggested to obtain the approximate posterior CPS. On this basis, the SI_CPSF algorithm, as an executable algorithm, is formed to solve the generic nonlinear filtering problem. Third, the approximate error between the CPSF framework and the recursive Bayesian filter is analyzed theoretically. The consistency of the CPSF framework to the recursive Bayesian filter is proved. Finally, the performances of the SI_CPSF algorithm, including robustness, accuracy and efficiency, are evaluated using four representative simulation experiments. The simulation results showed that SI_CSPF requires far less samples than particle filter (PF) under the same accuracy. Its computation is on average one order of magnitude less than that of the PF. The robustness of the proposed algorithm is also evaluated in the simulations.

Coexistence in multi-use landscape: linking human activities with functional traits of wild mammals in southern India

ContextHuman presence and land-use activities influence habitat use and activity of species. It is crucial to study the ecological and anthropogenic determinants that drive these relationships.ObjectivesWe investigated the effects of land-use change and human presence on the habitat use of seven mammalian species and on their activity patterns.MethodsWe conducted a camera-trapping survey in Ballari district, India. We deployed camera-traps at 1457 sampling grid cells across Protected Areas (PAs), Reserved Forests (RFs), and adjoining fringe mosaic farmlands (FMFLs). We assessed the influence of 19 covariates related to habitat features, landscape-level human presence, climate, and local-level human presence on species habitat use. We examined the diel activity patterns and activity levels of mammals within different land-use types.ResultsFour species exhibited a net negative response to increasing human presence, while three species exhibited positive responses. Respectively, proximity to farmlands, and settlements positively influenced habitat use by Panthera pardus, and Lepus nigricollis,. Increasing direct human presence negatively affected Sus scrofa habitat use. Large body-sized species exhibited higher diurnal activity in PA/RF compared to FMFL while smaller-sized (≤ 2.2 kg) species maintained similar activity patterns across land-use types. Generalist meso-predators exhibited more diurnal activity in RF/FMFL.ConclusionsEcological and life history traits were strong predictors of species responses to increasing human presence which favoured smaller and generalist species. These insights into the complex mechanisms through which human activities shape mammal communities are important landscape level perspectives on conservation.

Harnessing Hyper-Accumulator Plants for an Innovative Green Technology to Recover Nickle from Serpentine Soils in Sri Lanka

Phyto-mining is an environmentally sound green technique that has been employed for metal recovery from low-grade sources using hyperaccumulator plants. Due to its economic viability and wide social acceptance, this technology has been favoured over many conventional mining applications. Nickel (Ni) phyto-mining was introduced to the world with the identification of many Ni hyperaccumulator species grown in Ni-enriched metalliferous soils like serpentine, where they accumulate more than 1,000 mg/kg of Ni. Along the boundary of Highland and Vijayan Complexes in Sri Lanka, Ginigalpelessa, Indikolapelessa, and Ussangoda serpentinite deposits are closely located by showing similar geographical features. According to previous studies, these deposits contain a high concentration of Ni in the soil causing the evolution of serpentinite flora unique to these areas. This study aimed to identify the diversity and the presence of hyperaccumulator species that can be employed in Ni phyto-mining. More importantly, the study focused on identifying annual or biannual crop species with short lifespans that can be used in phyto-mining. The plant diversity and abundance analysis were carried out using a grid sampling method (3×3 m), and the highly abundant species were harvested and subjected to aqua regia digestion to determine hyperaccumulator levels using Inductively Coupled Plasma Mass Spectrometer (ICP-MS). Ginigalpelessa and Indikolapelessa showed similar diversity, abundant with grass and small herb species, such as Apluda mutica, Imperata cylindrica, Eupatorium odoratum, and Tephrosia villosa, where the A. mutica was dominant. In addition, plants like Morinda tinctoria, Vitex pinnata, and Azadirachta indica were dispersed throughout the deposit. Ussangoda deposit exhibited few herb species, including Hybanthus enneaspermus, Evolvulus alsinoides, and Desmodium triflorum, and localized thick vegetation patches. Diversity analysis of all the deposits showed very low plant diversity, ranging from 0.1 to 1.3. Moreover, the areas containing high Ni concentrations (6,000- 10,000 mg/kg) were abundant with stunted trees and grass species. The present study identified the Apluda grass species as a potential Ni hyperaccumulator for phytomining since it accumulates a high concentration of Ni (900-2,800 mg/kg) and can be grown biannually. Even though the grass species was dominant in both Ginigalpelessa and Indikolapelessa deposits, it was not observed in Ussangoda deposit, probably due to the arid climate and the deposition of salt spray from the coast. The findings of this study would be promising to establish phytomining in serpentine soils to support the increasing demand for Ni in the global market, especially for clean energy transition.
 
 Keywords: Low-grade soils, Clean energy transition, Ni phyto-mining, Serpentinite deposits
 
 Financial assistance from SR/LT/2021/15 grant from University of Moratuwa is acknowledged

AI-Based Prediction of Carrot Yield and Quality on Tropical Agriculture

The adoption of artificial intelligence tools can improve production efficiency in the agroindustry. Our objective was to perform the predictive modeling of carrot yield and quality. The crop was grown in two commercial areas during the summer season in Brazil. The root samples were taken at 200 points with a 30 × 30 m sampling grid at 82 and 116 days after sowing in both areas. The total fresh biomass, aerial part, and root biometry were quantified for previous crop harvesting to measure yield. The quality of the roots was assessed by sub-sampling three carrots by the concentration of total soluble solids (°Brix) and firmness in the laboratory. Vegetation indices were extracted from satellite imagery. The most important variables for the predictive models were selected by principal component analysis and submitted to the Artificial Neural Network (ANN), Random Forest (RF), and Multiple Linear Regression (MLR) algorithms. SAVI and NDVI indices stood out as predictors of crop yield, and the results from the ANN (R2 = 0.68) were superior to the RF (R2 = 0.67) and MLR (R2 = 0.61) models. Carrot quality cannot be modeled by the predictive models in this study; however, it should be explored in future research, including other crop variables.

Enhanced Global Convolution Networks for Defect Detection in Full Wavefield Imaging

Non-Destructive Evaluation (NDE) techniques based on ultrasonic guided waves can provide relevant information about the health status of an inspected structure. The analysis of the wavefields recorded through Scanning Laser Doppler Vibrometers (SLDVs) has been used successfully in NDE systems to detect damages in planar isotropic materials, such as interlaminar fractures in composites. However, the high spatial point resolution required to ensure an accurate localization and quantification of defects typically implies a time-consuming acquisition process, which limits the applicability of such approaches. In this work, we exploit the potential of convolutional neural networks (CNNs) to perform segmentation on full wavefield images, obtained by reconstruction from a low resolution grid having a spatial sampling rate below the Nyquist frequency, for the purpose of detecting and localising delaminations in carbon fibre reinforced plastic (CFRP) plates. In particular, we trained an improved version of Global Convolutional Networks (GCN) using a public dataset containing 475 simulated cases of full wavefield Lamb waves propagation in a CFRP plate, generated by an actuator with a carrier frequency of 50 kHz. We adopted channel and spatial attention mechanisms to improve the accuracy of the networks and applied our method on (i) the resulting image given by the root mean square value (RMS) in time for each spatial position and on (ii) the 3-dimensional animation representing the full wavefield propagation. Our networks show the ability to accurately locate target damages with a spatial resolution 8 times higher than the dimension of the adopted sampling grid, achieving an Intersection over Union score equal to 0.78 with a number of scanning point more than 60 times lower than the number of pixels in the output segmentation mask.

Accurate Fourier-space statistics for line intensity mapping: Cartesian grid sampling without aliased power

ABSTRACT Estimators for n-point clustering statistics in Fourier-space demand that modern surveys of large-scale structure be transformed to Cartesian coordinates to perform Fast Fourier Transforms (FFTs). In this work, we explore this transformation in the context of pixelized line intensity maps (LIM), highlighting potential biasing effects on power-spectrum measurements. Current analyses often avoid a complete resampling of the data by approximating survey geometry as rectangular in Cartesian space, an increasingly inaccurate assumption for modern wide-sky surveys. Our simulations of a $20\, {\times }\, 20\, \text{deg}^2$ 21 cm LIM survey at $0.34\, {\lt }\, z\, {\lt }\, 0.54$ show this assumption biases power-spectrum measurements by ${\gt }\, 20~{{\ \rm per\ cent}}$ across all scales. We therefore present a more robust framework for regridding the voxel intensities on to a 3D FFT field by coordinate transforming large numbers of Monte-Carlo sampling particles. Whilst this unbiases power-spectrum measurements on large scales, smaller scale discrepancies remain, caused by structure smoothing and aliasing from separations unresolved by the grid. To correct these effects, we introduce modelling techniques, higher order particle assignments, and interlaced FFT grids to suppress the aliased power. Using a piecewise cubic spline (PCS) particle assignment and an interlaced FFT field, we achieve sub-per cent accuracy up to 80 per cent of the Nyquist frequency for our 21 cm LIM simulations. We find a more subtle hierarchical improvement in results for higher order assignment schemes, relative to the gains made for galaxy surveys, which we attribute to the extra complexity in LIM from additional discretizing steps. python code accompanying this paper is available at github.com/stevecunnington/gridimp.