Effects Of Different Sampling Strategies Research Articles

Improved deep learning segmentation of outdoor point clouds with different sampling strategies and using intensities

Abstract The rapid growth of outdoor digital twin data sets and advancements in 3D data acquisition technology have sparked interest in improving segmentation performance using deep learning. This research aims to analyze and evaluate different sampling strategies and optimization techniques while exploring the intensity information of outdoor point cloud data. Two sampling strategies, random and stratified sampling, are employed to divide a limited data set. Additionally, the data set is divided into point cloud data with and without intensity. The PointNet++ model is used to segment the point cloud data into two classes, vegetation and structure. The results indicate that stratified sampling outperforms random sampling, yielding a considerable improvement in mean intersection over union scores of up to 10%. Interestingly, the inclusion of intensity information in the data set does not universally enhance performance. Although the use of intensity improves the performance of random sampling, it does not benefit stratified sampling. This research provides insights into the effectiveness of different sampling strategies for outdoor point cloud data segmentation. The findings can contribute to the development of optimized approaches to improving segmentation accuracy in outdoor digital twin applications using deep learning techniques.

Effects of different sampling strategies for unburned label selection in machine learning modelling of wildfire occurrence probability

The selection of unburned labels is a crucial step in machine learning modelling of wildfire occurrence probability. However, the effect of different sampling strategies on the performance of machine learning methods has not yet been thoroughly investigated. Additionally, whether the ratio of burned labels to unburned labels should be balanced or imbalanced remains a controversial issue. To address these gaps in the literature, we examined the effects of four broadly used sampling strategies for unburned label selection: (1) random selection in the unburned areas, (2) selection of areas with only one fire event, (3) selection of barren areas, and (4) selection of areas determined by the semi-variogram geostatistical technique. The effect of the balanced and imbalanced ratio between burned and unburned labels was also investigated. The random forest (RF) method explored the relationships between historical wildfires that occurred over the period between 2001 and 2020 in Yunnan Province, China, and climate, topography, fuel and anthropogenic variables. Multiple metrics demonstrated that the random selection of the unburned labels from the unburned areas with an imbalanced dataset outperformed the other three sampling strategies. Thus, we recommend this strategy to produce the required datasets for machine learning modelling of wildfire occurrence probability.

Effects of sampling strategy in rivers on load estimation for Nitrate-Nitrogen and total Phosphorus in a lowland agricultural area

The transport of nutrients into water bodies is one of the main causes of water eutrophication. It is therefore important to estimate the loads of nutrients. Discharge and nutrient concentrations are the fundamental elements to estimate the loads of nutrients, the latter can be affected by sampling strategies. As conducting sampling campaign and laboratory analysis are both expensive, it is necessary to find the best effective sampling strategy. The aim of this paper is to show how autocorrelation and standard statistical methods can be used to test the effects of different sampling strategies on the nutrient load estimation and to find the optimal sampling strategy. The data set in this study is from the 50 km² Kielstau catchment, a UNESCO demo site for ecohydrology in Northern Germany and consists of 14 years daily values of climate, hydrology, and water quality from 2006 to 2019. We calculated the autocorrelation (AC) of discharge (Q), precipitation, Nitrate-Nitrogen (NO3-N) and total Phosphorus (Ptot). Then we tested the effects of sampling intervals from 7 to 56 days (1–8 weeks) on the nutrient loads. Our results showed a high AC of Q and NO3-N for a long period, but the AC of Ptot and precipitation decreased very fast. An increase of the sampling interval (less frequent) increased the error of estimating the concentrations and loads. Consequently, we recommend that (1) the optimal sampling strategy for nutrient load estimation in an agriculture-dominant catchment should be continuously monitoring discharge combined with periodic grabbed samples; (2) the sampling frequency for NO3-N is suggested to be monthly (every 28 days) and for Ptot weekly (every 7 days). The information will help those tasked with catchment monitoring to design appropriate sampling strategy to ensure adequate data for nutrients load estimation in lowland rivers.

Impact of Training Set Configurations for Differentiating Plantation Forest Genera with Sentinel-2 Imagery and Machine Learning

Forest plantations in South Africa impose genus-specific demands on limited soil moisture. Hence, plantation composition and distribution mapping is critical for water conservation planning. Genus maps are used to quantify the impact of post-harvest genus-exchange activities in the forestry sector. Collecting genus data using in situ methods is costly and time-consuming, especially when performed at regional or national scales. Although remotely sensed data and machine learning show potential for mapping genera at regional scales, the efficacy of such methods is highly dependent on the size and quality of the training data used to build the models. However, it is not known what sampling scheme (e.g., sample size, proportion per genus, and spatial distribution) is most effective to map forest genera over large and complex areas. Using Sentinel-2 imagery as inputs, this study evaluated the effects of different sampling strategies (e.g., even, uneven, and area-proportionate) for training the random forests machine learning classifier to differentiate between Acacia, Eucalyptus, and Pinus trees in South Africa. Sample size (s) was related to the number of input features (n) to better understand the potential impact of sample sparseness. The results show that an even sample with maximum size (100%, s~91n) produced the highest overall accuracy (76.3%). Although larger training set sizes (s > n) resulted in higher OAs, a saturation point was reached at s~64n.

Galaxy morphology classification using automated machine learning

In this paper, we apply five different machine learning algorithms to classify samples into four categories — spirals, ellipticals, mergers and stars (don’t know) using data from the Sloan Digital Sky Survey to assess the feasibility of using machine learning methods for future surveys. Classifying mergers as a separate class poses a challenge as this category is easily confused with both ellipticals and spirals, and as a result, most previous studies have not included mergers as a distinct morphological class. The dataset is highly imbalanced with the number of ellipticals/spirals being much larger than the number of stars/mergers, and this is another challenge we aim to address. Starting with 62 features, we perform principal component analysis and use the 25 most significant principal components as inputs to the machine learning models. We compare our results with the Galaxy Zoo labels and obtain an overall test accuracy of 98.2% and 97.5% using Artificial Neural Network and ExtraTrees respectively. However, ExtraTrees outperforms Neural Network in classifying mergers and stars. We also perform a parameter sensitivity test to compare the relative importance of different categories of features on the model’s performance. Finally, we address the class imbalance problem and examine the effects of different sampling strategies. Our results show that the use of a balanced dataset with a large number of training samples leads to high recall values for the minority classes, and that oversampling methods lead to better performance than undersampling techniques.

Sampling cores and sequencing depths affected the measurement of microbial diversity in soil quadrats

Due to the massive quantity and broad phylogeny, an accurate measurement of microbial diversity is highly challenging in soil ecosystems. Initially, the deviation caused by sampling should be adequately considered. Here, we attempted to uncover the effect of different sampling strategies on α diversity measurement of soil prokaryotes. Four 1 m2 sampling quadrats in a typical grassland were thoroughly surveyed through deep 16S rRNA gene sequencing (over 11 million reads per quadrat) with numerous replicates (33 soil sampling cores with total 141 replicates per quadrat). We found the difference in diversity was relatively small when pooling soil cores before and after DNA extraction and sequencing, but they were both superior to a non-pooling strategy. Pooling a small number of soil cores (i.e., 5 or 9) combined with several technical replicates is sufficient to estimate diversities for soil prokaryotes, and there is great flexibility in pooling original samples or data at different experimental steps. Additionally, the distribution of local α diversity varies with sampling core number, sequencing depth, and abundance distribution of the community, especially for high orders of Hill diversity index (i.e., Shannon entropy and inverse Simpson index). For each grassland soil quadrat (1 m2), retaining 100,000 reads after taxonomic clustering might be a realistic option, as these number of reads can efficiently cover the majority of common species in this area. Our findings provide important guidance for soil sampling strategy, and the general results can serve as a basis for further studies.

Sampling strategies in the assessment of long‐term exposures to toxic substances in air

AbstractThe decision to mitigate exposures from vapor intrusion (VI) is typically based on limited data from 24‐hour air samples. It is well documented that these data do not accurately represent long‐term average exposures linked to adverse health effects. Limited decision guidance is currently available to determine the most appropriate sampling strategy, considering the cost of sampling alternatives along with the economic consequences of exposure‐related health effects. We present a decision model that introduces economic and statistical considerations in evaluating alternative VI sampling methods. The model characterizes the best sampling method by factoring economic and health consequences of exposure, the variability of exposure, the cost of sampling and mitigation, and the likelihood of false‐negatives and false‐positives. Decision‐makers can use results to select the sample size that maximizes net benefit. Conceptual and mathematical models are presented linking biological, statistical, and economic considerations to assess the cost and effectiveness of different sampling strategies. The model relates an average exposure concentration, determined statistically, to abatement costs and to the monetary value of health deterioration. The value of the information provided by different strategies is calculated and used to select the optimum sampling method. Simulations show that longer‐term sampling methods tend to be more accurate and cost‐effective than short‐term samples. The ideal sampling strategy shows significant seasonal variation (it is typically optimal to use longer samples in the winter) and also varies significantly with the stringency of regulatory standards. Longer‐term sample collection provides a more accurate representation of average VI exposure and reduces the likelihood of type I and type II errors. This reduces expected costs of mitigation and exposure (e.g., health consequences, legal and regulatory penalties), which in some cases can be quite significant. The model herein shows how these savings are balanced against the additional costs of longer‐term sampling.

Phylogenomic species delimitation in microendemic frogs of the Brazilian Atlantic Forest

The advent of next-generation sequencing allows researchers to use large-scale datasets for species delimitation analyses, yet one can envision an inflection point where the added accuracy of including more loci does not offset the increased computational burden. One alternative to including all loci could be to prioritize the analysis of loci for which there is an expectation of high informativeness. Here, we explore the issue of species delimitation and locus selection with montane species from two anuran genera that have been isolated in sky islands across the southern Brazilian Atlantic Forest: Melanophryniscus (Bufonidae) and Brachycephalus (Brachycephalidae). To delimit species, we obtained genetic data using target enrichment of ultraconserved elements from 32 populations (13 for Melanophryniscus and 19 for Brachycephalus), and we were able to create datasets that included over 800 loci with no missing data. We ranked loci according to their number of parsimony-informative sites, and we performed species delimitation analyses using BPP with the most informative 10, 20, 40, 80, 160, 320, and 640 loci. We identified three types of phylogenetic node: nodes with either consistently high or low support regardless of the number of loci or their informativeness and nodes that were initially poorly supported where support became stronger as we included more data. When viewed across all sensitivity analyses, our results suggest that the current species richness in both genera is likely underestimated. In addition, our results show the effects of different sampling strategies on species delimitation using phylogenomic datasets.

Evaluation of methods to sample fecal indicator bacteria in foreshore sand and pore water at freshwater beaches

Fecal indicator bacteria (FIB) are known to accumulate in foreshore beach sand and pore water (referred to as foreshore reservoir) where they act as a non-point source for contaminating adjacent surface waters. While guidelines exist for sampling surface waters at recreational beaches, there is no widely-accepted method to collect sand/sediment or pore water samples for FIB enumeration. The effect of different sampling strategies in quantifying the abundance of FIB in the foreshore reservoir is unclear. Sampling was conducted at six freshwater beaches with different sand types to evaluate sampling methods for characterizing the abundance of E. coli in the foreshore reservoir as well as the partitioning of E. coli between different components in the foreshore reservoir (pore water, saturated sand, unsaturated sand). Methods were evaluated for collection of pore water (drive point, shovel, and careful excavation), unsaturated sand (top 1 cm, top 5 cm), and saturated sand (sediment core, shovel, and careful excavation). Ankle-depth surface water samples were also collected for comparison. Pore water sampled with a shovel resulted in the highest observed E. coli concentrations (only statistically significant at fine sand beaches) and lowest variability compared to other sampling methods. Collection of the top 1 cm of unsaturated sand resulted in higher and more variable concentrations than the top 5 cm of sand. There were no statistical differences in E. coli concentrations when using different methods to sample the saturated sand. Overall, the unsaturated sand had the highest amount of E. coli when compared to saturated sand and pore water (considered on a bulk volumetric basis). The findings presented will help determine the appropriate sampling strategy for characterizing FIB abundance in the foreshore reservoir as a means of predicting its potential impact on nearshore surface water quality and public health risk.

A Comparative Study of Sampling Analysis in the Scene Classification of Optical High-Spatial Resolution Remote Sensing Imagery

Scene classification, which consists of assigning images with semantic labels by exploiting the local spatial arrangements and structural patterns inside tiled regions, is a key problem in the automatic interpretation of optical high-spatial resolution remote sensing imagery. Many state-of-the-art methods, e.g., the bag-of-visual-words model and its variants, the topic models and unsupervised feature learning-based approaches, share similar procedures: patch sampling, feature learning and classification. Patch sampling is the first and a key procedure, and it has a considerable influence on the results. In the literature, many different sampling strategies have been used, e.g., random sampling and saliency-based sampling. However, the sampling strategy that is most suitable for the scene classification of optical high-spatial resolution remote sensing images remains unclear. In this paper, we comparatively study the effects of different sampling strategies under the scenario of scene classification of optical high-spatial resolution remote sensing images. We divide the existing sampling methods into two types: random sampling and saliency-based sampling. Here, we consider the commonly-used grid sampling to be a specific type of random sampling method, and the saliency-based sampling consists of keypoint-based sampling and salient region-based sampling. To compare their performances, we rely on a standard bag-of-visual-words model to learn the global features for testing because of its simplicity, robustness and efficiency. In addition, we conduct experiments using a Fisher kernel framework to validate our conclusions. The experimental results obtained on two commonly-used datasets using different feature learning methods show that random sampling can provide comparable and even better performance than all of the saliency-based strategies.

A Probabilistic Analysis of Sparse Coded Feature Pooling and Its Application for Image Retrieval.

Feature coding and pooling as a key component of image retrieval have been widely studied over the past several years. Recently sparse coding with max-pooling is regarded as the state-of-the-art for image classification. However there is no comprehensive study concerning the application of sparse coding for image retrieval. In this paper, we first analyze the effects of different sampling strategies for image retrieval, then we discuss feature pooling strategies on image retrieval performance with a probabilistic explanation in the context of sparse coding framework, and propose a modified sum pooling procedure which can improve the retrieval accuracy significantly. Further we apply sparse coding method to aggregate multiple types of features for large-scale image retrieval. Extensive experiments on commonly-used evaluation datasets demonstrate that our final compact image representation improves the retrieval accuracy significantly.

Effect of different sampling strategies for a single geographic region in Yemen on standard genetic analyses of mitochondrial DNA sequence data

Collection of biological samples is the foundation of genetic studies ranging from estimation of genetic diversity to reconstruction of population history. Sample collections are intended to accurately represent the genetic, biological, ecological, cultural, geographic, and/or linguistic diversity of a particular region or population by providing a small, but representative, set of samples. In this study, we analyze human mitochondrial DNA variation in samples collected using four different sampling strategies to represent the same geographic region. Specifically, samples were collected from a village, a rural area, a regional clinic, and a national university in the governorate of Dhamar in Yemen. All samples were assayed for mitochondrial hypervariable region I DNA sequence variation and data were subjected to standard molecular genetic analyses. Our results suggest that analyses in which individual DNA sequences are explicitly compared or evaluated, e.g. phylogenetic and network analyses, may be more sensitive to sample collection design than analyses in which data are averaged across individuals or are analyzed more indirectly, e.g. summary statistics.

Surface Fuel Sampling Strategies: Linking Fuel Measurements and Fire Effects

Abstract We assessed the effectiveness of different sampling strategies in linking fine fuel load and crown scorch of ashe (Juniperus ashei) and redberry juniper (J. pinchotii) for prescribed fires conducted in wet and dry periods of the growing season on the Edwards Plateau, Texas, USA. Our aim was to determine if spatial and temporal variation in crown scorch was best predicted by estimates of fuel load sampled with spatially explicit, multiscale sampling strategies or with traditional, simple random sampling of fuel load. We found that multiscale sampling of fuel load underneath and adjacent to juniper crowns was more effective than simple random sampling in predicting crown scorch for the 14 fires conducted in the wet period and the five conducted in the dry period. The type of sampling strategy employed was critical in relating fuel load to crown scorch during the wet period. Percent crown volume scorched ranged from 0% to 100% in these conditions. In contrast, the type of sampling strategy was less ...

To pool or not to pool, from whether to when: applications of pooling to biospecimens subject to a limit of detection

Pooling of biological specimens has been utilised as a cost-efficient sampling strategy, but cost is not the unique limiting factor in biomarker development and evaluation. We examine the effect of different sampling strategies of biospecimens for exposure assessment that cannot be detected below a detection threshold (DT). The paper compares use of pooled samples to a randomly selected sample from a cohort in order to evaluate the efficiency of parameter estimates. The proposed approach shows that a pooling design is more efficient than a random sample strategy under certain circumstances. Moreover, because pooling minimises the amount of information lost below the DT, the use of pooled data is preferable (in a context of a parametric estimation) to using all available individual measurements, for certain values of the DT. We propose a combined design, which applies pooled and unpooled biospecimens, in order to capture the strengths of the different sampling strategies and overcome instrument limitations (i.e. DT). Several Monte Carlo simulations and an example based on actual biomarker data illustrate the results of the article.

Sample Size Computations for PK/PD Population Models

We describe an accurate, yet simple and fast sample size computation method for hypothesis testing in population PK/PD studies. We use a first order approximation to the nonlinear mixed effects model and chi-square distributed Wald statistic to compute the minimum sample size to achieve given degree of power in rejecting a null hypothesis in population PK/PD studies. The method is an extension of Rochon's sample size computation method for repeated measurement experiments. We compute sample sizes for PK and PK/PD models with different conditions, and use Monte Carlo simulation to show that the computed sample size retrieves the required power. We also show the effect of different sampling strategies, such as minimal, i.e., as many observations per individual as parameters in the model, and intensive on sample size. The proposed sample size computation method can produce estimates of minimum sample size to achieve the desired power in hypothesis testing in a greatly reduced time than currently available simulation-based methods. The method is rapid and efficient for sample size computation in population PK/PD study using nonlinear mixed effect models. The method is general and can accommodate any type of hierarchical models. Simulation results suggest that intensive sampling allows the reduction of the number of patients enrolled in a clinical study.

A three dimensional model of Cryptosporidium dynamics in lakes and reservoirs: A new tool for risk management

A model of Cryptosporidium oocyst dynamics for lakes and reservoirs is presented. The model consists of a module that simulates oocyst inactivation, resuspension, settling and aggregation onto particles. This module was coupled to the three‐dimensional Estuary Lake and Coastal Ocean Model (ELCOM), which was used to simulate lake hydrodynamics in addition to oocyst advection and turbulent diffusion. A field experiment that tracked the passage of a flood inflow throughout Myponga Reservoir, South Australia, was used to validate the coupled model. The model accurately captured the thermal dynamics, and the spatial and temporal distribution of different inorganic particle size classes and oocysts. The model and data indicate that oocysts do not readily attach to inorganic particles as other researchers have suggested but settle as free‐floating oocysts according to Stoke's sedimentation dynamics. The reduction in oocysts between the inflow and the offtake due to settling is therefore not as significant as previously thought. The potential for inactivation was also found to be small relative to the timescales for transport. The model is a useful tool to examine oocyst dynamics in lakes and reservoirs, to consider risk management assessments of different scenarios, and to assess the effectiveness of different sampling strategies.

Effects of different sampling strategies on predictions of blood cyclosporine concentrations in haematological patients with multidrug resistance by bayesian and non-linear least squares methods

The Bayesian method (BM) can use previous information for the optimization of dosage regimen. However, Bayes' law remains true when the parameters are obtained from the infinite population. Therefore a bias might exist in the previous information and affect BM predictive performance. To overcome this shortcoming, the blood drug concentration of a patient can be used to individualize his pharmacokinetic parameters. Until now, at least two sampling strategies, i.e. steady-state and non-steady-state sampling strategies, have been developed to individualize and predict blood drug concentration. In the present study we used five sampling strategies: (1) all samples; (2) post-infusion samples; (3) during-infusion samples; (4) samples within 95% confidence interval/interquartile range of a steady-state concentration; (5) the sample of the mean/median at the mid-time-point of a steady-state to individualize and predict blood cyclosporine concentrations in haematological patients with multidrug resistance. We investigated the effects of different sampling strategies on BM and the nonlinear least squared method (NLLSM) predictive performances. The results showed that BM predictive performance was better than NLLSM. But the results did not prove that the steady-state sampling strategies were superior to the non-steady-state ones.

A study of sampling strategies for estimating growth parameters in fish populations

The effects of different sampling strategies for age and growth studies and for the estimation of growth parameters from size-at-age data were investigated using simulated age-structured populations of two fish species representing different life history characteristics. One species type was a small pelagic with relatively fast growth, small size, high mortality, low longevity and is representative of r-strategists. The second case was a K-strate-gist (lutjanid/sparid) type with relatively larger size, slower growth, greater longevity and lower mortality. Three different sampling strategies were evaluated: (1) simple random sampling; (2) proportional allocation; (3) fixed allocation. Overall, there is no best sampling strategy in all situations. Proportional allocation tends to show smaller parameter variability. Overall, there seems to be no advantage in fixed allocation in relation to proportional allocation or even simple random sampling.

Estimating the Frequencies of Genetically Distinct Classes of Individuals in Hybridized Populations

Hybridization and interspecific recombination result in the formation of novel morphologies and associations of molecular markers. Thus, in the past, the existence of hybridization and introgression has been inferred by identifying individuals exhibiting some defined hybrid or backcross phenotypes or genotypes. This approach can be unreliable because it may under- or overestimate the frequency of different classes of individuals in a hybridized population. Also, this method provides no formal framework for testing a null hypothesis that the frequency of a particular class of hybrids or introgressants is equal to zero. A model is presented which uses the method of maximum likelihood and codominant molecular markers segregating for parental species specific alleles to estimate the frequencies of six classes of individuals that may occur in a hybridized population. Estimates of the frequency of parental species, hybrid, and introgressed individuals are unbiased if model assumptions are met. If model assumptions are violated, estimates of the frequency of introgression are conservative. Monte Carlo computer simulations are used to examine the bias and variance of the model estimates and to evaluate the effects of different sampling strategies as the number of loci with unique alleles, the frequencies of these alleles, and the number of putative hybrid individuals sampled are varied. Simulations were also used to examine the accuracy of delta and bootstrap methods of obtaining true variances of the model estimates. This approach provides evolutionary geneticists with a simple method to quantify the occurrence and extent of hybridization and introgression in natural populations.

Hydrologic sampling — A characterization in terms of rainfall and basin properties

This paper considers the sampling of rainfall and discharge processes both in time and in space and links the sampling problem to basin and rainfall characteristics. The effectiveness of different sampling strategies is measured by the variance of the error in estimating either total or peak of streamflow from a single storm event. This is related to the rainfall and basin rainfall-discharge properties through parameterizations of these processes. Rainfall is modeled as a collection of rain cells which occur randomly in space and time and has parameters which define the probability of occurrence of rain cells in space and time and the spread of rainfall due to a cell. Discharge from rainfall is parameterized in terms of the fluvial geomorphology of the basin. Linear filtering techniques are used to compute the variance of the estimation error for different sampling strategies. Sampling strategies are defined by the number of rain gages, rainfall sampling interval and discharge measurement interval. The results can be used in hydrologic network design to assess the effectiveness of different sampling options.