Multivariate estimation of the limit of detection by orthogonal partial least squares in temperature-modulated MOX sensors

Gas Sensor Drift Compensation Using Semi-Supervised Ensemble Classifiers with Multi-Level Features and Center Loss

The drift compensation of gas sensors is a significant and challenging issue in the field of electronic noses (E-nose). Compensating sensor drift has a great benefit in improving the performance of E-nose systems. However, conventional methods often perform poorly due to complex data relationships before and after drifting, or require label information for both nondrift (source data) and drift data (target data) to enhance performance, which is hard to achieve and even unrealistic. In this study, we propose a semisupervised domain adaptive convolutional neural network (CNN) based on ensemble classifiers of multilevel features, pretraining, and center loss to tackle the drift problem. The main idea is to make full use of multilevel features extracted from the network and apply Hilbert space's maximum mean discrepancy (MMD) to evaluate the domain similarity of the features at different levels. Then the corresponding MMD is used as a weight to achieve the weighted fusion of predictions in the classifier ensemble module, so as to obtain a more reliable result. Furthermore, to optimize training, MMD is used as a loss for pretraining to help feature extractors learn more robust and common features in two domains. Center loss is also applied to achieve more focused learning for features of the same class. The results on two data sets demonstrate the effectiveness of our method. The average classification accuracies under different settings reach 76.06% (long-drift) and 82.07% (short-drift), respectively, and the average R2 score reaches 0.804 in the regression task, which has significant improvements compared with several conventional methods. Our work provides an effective and reliable method at the algorithm level to solve the drift compensation problem of gas sensors.

Quantification of Lanthanides on a PMMA Microfluidic Device with Three Optical Pathlengths Using PCR of UV-Visible, NIR, and Raman Spectroscopy

Microfluidic devices (MFDs) offer customizable, low-cost, and low-waste platforms for performing chemical analyses. Optical spectroscopy techniques provide nondestructive monitoring of small sample volumes within microfluidic channels. Optical spectroscopy can probe speciation, oxidation state, and concentration of analytes as well as detect counterions and provide information about matrix composition. Here, ultraviolet-visible (UV-vis) absorbance, near-infrared (NIR) absorbance, and Raman spectroscopy are utilized on a custom poly(methyl methacrylate) (PMMA) MFD for the detection of three lanthanide nitrates in solution. Absorbance spectroscopies are conducted across three pathlengths using three portions of a contiguous channel within the MFD. Univariate and chemometric multivariate modeling, specifically Beer's law regression and principal component regression (PCR), respectively, are utilized to quantify the three lanthanides and the nitrate counterion. Models are composed of spectra from one or multiple pathlengths. Models are also constructed from multiblock spectra composed of UV-vis, NIR, and Raman spectra at one or multiple pathlengths. Root-mean-square errors (RMSE), limit of detection (LOD), and residual predictive deviation (RPD) values are compared for univariate, multivariate, multi-pathlength, and multiblock models. Univariate modeling produces acceptable results for analytes with a simple signal, such as samarium cations, producing an LOD of 5.49 mM. Multivariate and multiblock models produce enhanced quantification for analytes that experience spectral overlap and interfering nonanalyte signals, such as holmium, which had an LOD reduction from 7.21 mM for the univariate model down to 3.96 mM for the multiblock model. Multi-pathlength models are developed that maintain model errors in line with single-pathlength models. Multi-pathlength models have RPDs from 9.18 to 46.4, while incorporating absorbance spectra collected at optical paths of up to 10-fold difference in length.

Methods for Estimating the Detection and Quantification Limits of Key Substances in Beer Maturation with Electronic Noses

To evaluate the suitability of an analytical instrument, essential figures of merit such as the limit of detection (LOD) and the limit of quantification (LOQ) can be employed. However, as the definitions k nown in the literature are mostly applicable to one signal per sample, estimating the LOD for substances with instruments yielding multidimensional results like electronic noses (eNoses) is still challenging. In this paper, we will compare and present different approaches to estimate the LOD for eNoses by employing commonly used multivariate data analysis and regression techniques, including principal component analysis (PCA), principal component regression (PCR), as well as partial least squares regression (PLSR). These methods could subsequently be used to assess the suitability of eNoses to help control and steer processes where volatiles are key process parameters. As a use case, we determined the LODs for key compounds involved in beer maturation, namely acetaldehyde, diacetyl, dimethyl sulfide, ethyl acetate, isobutanol, and 2-phenylethanol, and discussed the suitability of our eNose for that dertermination process. The results of the methods performed demonstrated differences of up to a factor of eight. For diacetyl, the LOD and the LOQ were sufficiently low to suggest potential for monitoring via eNose.

PVS-GEN: Systematic Approach for Universal Synthetic Data Generation Involving Parameterization, Verification, and Segmentation

Synthetic data generation addresses the challenges of obtaining extensive empirical datasets, offering benefits such as cost-effectiveness, time efficiency, and robust model development. Nonetheless, synthetic data-generation methodologies still encounter significant difficulties, including a lack of standardized metrics for modeling different data types and comparing generated results. This study introduces PVS-GEN, an automated, general-purpose process for synthetic data generation and verification. The PVS-GEN method parameterizes time-series data with minimal human intervention and verifies model construction using a specific metric derived from extracted parameters. For complex data, the process iteratively segments the empirical dataset until an extracted parameter can reproduce synthetic data that reflects the empirical characteristics, irrespective of the sensor data type. Moreover, we introduce the PoR metric to quantify the quality of the generated data by evaluating its time-series characteristics. Consequently, the proposed method can automatically generate diverse time-series data that covers a wide range of sensor types. We compared PVS-GEN with existing synthetic data-generation methodologies, and PVS-GEN demonstrated a superior performance. It generated data with a similarity of up to 37.1% across multiple data types and by 19.6% on average using the proposed metric, irrespective of the data type.

Gas Sensor Array Fault Diagnosis Based on Multi-Dimensional Fusion, an Attention Mechanism, and Multi-Task Learning

With the development of gas sensor arrays and computational technology, machine olfactory systems have been widely used in environmental monitoring, medical diagnosis, and other fields. The reliable and stable operation of gas sensing systems depends heavily on the accuracy of the sensors outputs. Therefore, the realization of accurate gas sensor array fault diagnosis is essential to monitor the working status of sensor arrays and ensure the normal operation of the whole system. The existing methods extract features from a single dimension and require the separate training of models for multiple diagnosis tasks, which limits diagnostic accuracy and efficiency. To address these limitations, for this study, a novel fault diagnosis network based on multi-dimensional feature fusion, an attention mechanism, and multi-task learning, MAM-Net, was developed and applied to gas sensor arrays. First, feature fusion models were applied to extract deep and comprehensive features from the original data in multiple dimensions. A residual network equipped with convolutional block attention modules and a Bi-LSTM network were designed for two-dimensional and one-dimensional signals to capture spatial and temporal features simultaneously. Subsequently, a concatenation layer was constructed using feature stitching to integrate the fault details of different dimensions and avoid ignoring useful information. Finally, a multi-task learning module was designed for the parallel learning of the sensor fault diagnosis to effectively improve the diagnosis capability. The experimental results derived from using the proposed framework on gas sensor datasets across different amounts of data, balanced and unbalanced datasets, and different experimental settings show that the proposed framework outperforms the other available methods and demonstrates good recognition accuracy and robustness.

Environmental Engineering Applications of Electronic Nose Systems Based on MOX Gas Sensors

Nowadays, the electronic nose (e-nose) has gained a huge amount of attention due to its ability to detect and differentiate mixtures of various gases and odors using a limited number of sensors. Its applications in the environmental fields include analysis of the parameters for environmental control, process control, and confirming the efficiency of the odor-control systems. The e-nose has been developed by mimicking the olfactory system of mammals. This paper investigates e-noses and their sensors for the detection of environmental contaminants. Among different types of gas chemical sensors, metal oxide semiconductor sensors (MOXs) can be used for the detection of volatile compounds in air at ppm and sub-ppm levels. In this regard, the advantages and disadvantages of MOX sensors and the solutions to solve the problems arising upon these sensors' applications are addressed, and the research works in the field of environmental contamination monitoring are overviewed. These studies have revealed the suitability of e-noses for most of the reported applications, especially when the tools were specifically developed for that application, e.g., in the facilities of water and wastewater management systems. As a general rule, the literature review discusses the aspects related to various applications as well as the development of effective solutions. However, the main limitation in the expansion of the use of e-noses as an environmental monitoring tool is their complexity and lack of specific standards, which can be corrected through appropriate data processing methods applications.

Estimating the Analytical Performance of Raman Spectroscopy for Quantification of Active Ingredients in Human Stratum Corneum

Confocal Raman microscopy (CRM) has become a versatile technique that can be applied routinely to monitor skin penetration of active molecules. In the present study, CRM coupled to multivariate analysis (namely PLSR—partial least squares regression) is used for the quantitative measurement of an active ingredient (AI) applied to isolated (ex vivo) human stratum corneum (SC), using systematically varied doses of resorcinol, as model compound, and the performance is quantified according to key figures of merit defined by regulatory bodies (ICH, FDA, and EMA). A methodology is thus demonstrated to establish the limit of detection (LOD), precision, accuracy, sensitivity (SEN), and selectivity (SEL) of the technique, and the performance according to these key figures of merit is compared to that of similar established methodologies, based on studies available in literature. First, principal components analysis (PCA) was used to examine the variability within the spectral data set collected. Second, ratios calculated from the area under the curve (AUC) of characteristic resorcinol and proteins/lipids bands (1400–1500 cm⁻¹) were used to perform linear regression analysis of the Raman spectra. Third, cross-validated PLSR analysis was applied to perform quantitative analysis in the fingerprint region. The AUC results show clearly that the intensities of Raman features in the spectra collected are linearly correlated to resorcinol concentrations in the SC (R² = 0.999) despite a heterogeneity in the distribution of the active molecule in the samples. The Root Mean Square Error of Cross-Validation (RMSECV) (0.017 mg resorcinol/mg SC), The Root Mean Square of Prediction (RMSEP) (0.015 mg resorcinol/mg SC), and R² (0.971) demonstrate the reliability of the linear regression constructed, enabling accurate quantification of resorcinol. Furthermore, the results have enabled the determination, for the first time, of numerical criteria to estimate analytical performances of CRM, including LOD, precision using bias corrected mean square error prediction (BCMSEP), sensitivity, and selectivity, for quantification of the performance of the analytical technique. This is one step further towards demonstrating that Raman spectroscopy complies with international guidelines and to establishing the technique as a reference and approved tool for permeation studies.

Assessing over Time Performance of an eNose Composed of 16 Single-Type MOX Gas Sensors Applied to Classify Two Volatiles

This paper assesses the over time performance of a custom electronic nose (eNose) composed of an array of commercial low-cost and single-type miniature metal-oxide (MOX) semiconductor gas sensors. The eNose uses 16 BME680 versatile sensor devices, each including an embedded non-selective MOX gas sensor that was originally proposed to measure the total volatile organic compounds (TVOC) in the air. This custom eNose has been used previously to detect ethanol and acetone, obtaining initial promising classification results that worsened over time because of sensor drift. The current paper assesses the over time performance of different classification methods applied to process the information gathered from the eNose. The best classification results have been obtained when applying a linear discriminant analysis (LDA) to the normalized conductance of the sensing layer of the 16 MOX gas sensors available in the eNose. The LDA procedure by itself has reduced the influence of drift in the classification performance of this single-type eNose during an evaluation period of three months.

Metal loaded nano-carbon gas sensor array for pollutant detection. NANOTECHNOLOGY 2022;33:195501. [PMID: 35073524 DOI: 10.1088/1361-6528/ac4e43]

Many research works report a sensitive detection of a wide variety of gas species. However, their in-lab detection is usually performed by using single gases and, therefore, selectivity often remains an unsolved issue. This paper reports a four-sensor array employing different nano-carbon sensitive layers (bare graphene, SnO₂@Graphene, WO₃@Graphene, and Au@CNTs). The different gas-sensitive films were characterised via several techniques such as FESEM, TEM, and Raman. First, an extensive study was performed to detect isolated NO₂, CO₂, and NH₃molecules, unravelling the sensing mechanism at the operating temperatures applied. Besides, the effect of the ambient moisture was also evaluated. Afterwards, a model for target gas identification and concentration prediction was developed. Indeed, the sensor array was used in mixtures of NO₂and CO₂for studying the cross-sensitivity and developing a calibration model. As a result, the NO₂detection with different background levels of CO₂was achieved with anR²of 0.987 and an RMSE of about 22 ppb.

Simultaneous detection of glucose, triglycerides, and total cholesterol in whole blood by Fourier-Transform Raman spectroscopy

In this paper, a reagent-free simultaneous and direct detection method of three analytes in human blood based on Fourier-transform Raman (FT-Raman) spectroscopy with 1064 nm laser radiation was proposed for the first time. A total of 161 human blood samples were characterized by FT-Raman spectroscopy under the excitation laser source of 1064 nm. In order to achieve a robust regression model, the Nonlinear Iterative Partial Least Squares (NIPALS) with orthogonal signal correction (OSC) algorithm and sample set partition based on a joint x-y distance (SPXY) is used to establish multivariate calibration models. The root means square error of cross-validation (RMSECV), root mean square error of prediction (RMSEP), correlation coefficients (R²) and ratio of performance to deviation (RPD) were 0.34255 mg/dL, 0.3662 mg/dL, 0.99982 and 56.3524 for glucose, 0.33656 mg/dL, 0.75736 mg/dL, 0.99967 and 34.9169 for total cholesterol (TC), and 0.29956 mg/dL, 0.27469 mg/dL, 0.99998 and 173.5098 for triglycerides (TG), respectively. The analysis results showed that the proposed method could be able to accurately predict the concentration of glucose, TC and TG in blood. This method can instantaneous multi-component detection on whole blood.

On the Use of Pulsed UV or Visible Light Activated Gas Sensing of Reducing and Oxidising Species with WO3 and WS2 Nanomaterials

This paper presents a methodology to quantify oxidizing and reducing gases using n-type and p-type chemiresistive sensors, respectively. Low temperature sensor heating with pulsed UV or visible light modulation is used together with the application of the fast Fourier transform (FFT) to extract sensor response features. These features are further processed via principal component analysis (PCA) and principal component regression (PCR) for achieving gas discrimination and building concentration prediction models with R² values up to 98% and RMSE values as low as 5% for the total gas concentration range studied. UV and visible light were used to study the influence of the light wavelength in the prediction model performance. We demonstrate that n-type and p-type sensors need to be used together for achieving good quantification of oxidizing and reducing species, respectively, since the semiconductor type defines the prediction model’s effectiveness towards an oxidizing or reducing gas. The presented method reduces considerably the total time needed to quantify the gas concentration compared with the results obtained in a previous work. The use of visible light LEDs for performing pulsed light modulation enhances system performance and considerably reduces cost in comparison to previously reported UV light-based approaches.

Electrospun ZnO/Pd Nanofibers: CO Sensing and Humidity Effect

Variable air humidity affects the characteristics of semiconductor metal oxides, which complicates the reliable and reproducible determination of CO content in ambient air by resistive gas sensors. In this work, we determined the sensor properties of electrospun ZnO and ZnO/Pd nanofibers in the detection of CO in dry and humid air, and investigated the sensing mechanism. The microstructure of the samples, palladium content, and oxidation state, type, and concentration of surface groups were characterized using complementary techniques: X-ray fluorescent spectroscopy, XRD, high-resolution transmission electron microscopy (HRTEM), high angle annular dark field scanning transmission electron microscopy (HAADF-STEM), energy-dispersive X-ray (EDX) mapping, XPS, and FTIR spectroscopy. The sensor properties of ZnO and ZnO/Pd nanofibers were studied at 100-450 °C in the concentration range of 5-15 ppm CO in dry (RH₂₅ = 0%) and humid (RH₂₅ = 60%) air. It was found that under humid conditions, ZnO completely loses its sensitivity to CO, while ZnO/Pd retains a high sensor response. On the basis of in situ diffuse reflectance IR Fourier transform spectroscopy (DRIFTS) results, it was concluded that high sensor response of ZnO/Pd nanofibers in dry and humid air was due to the electronic sensitization effect, which was not influenced by humidity change.

Environmental chemical sensing using small drones: A review

Recent advances in miniaturization of chemical instrumentation and in low-cost small drones are catalyzing exponential growth in the use of such platforms for environmental chemical sensing applications. The versatility of chemically sensitive drones is reflected by their rapid adoption in scientific, industrial, and regulatory domains, such as in atmospheric research studies, industrial emission monitoring, and in enforcement of environmental regulations. As a result of this interdisciplinarity, progress to date has been reported across a broad spread of scientific and non-scientific databases, including scientific journals, press releases, company websites, and field reports. The aim of this paper is to assemble all of these pieces of information into a comprehensive, structured and updated review of the field of chemical sensing using small drones. We exhaustively review current and emerging applications of this technology, as well as sensing platforms and algorithms developed by research groups and companies for tasks such as gas concentration mapping, source localization, and flux estimation. We conclude with a discussion of the most pressing technological and regulatory limitations in current practice, and how these could be addressed by future research.

Fast, direct and in situ monitoring of lipid oxidation in an oil-in-water emulsion by near infrared spectroscopy

Lipid oxidation has implications on food, cosmetics and other fat containing products. Fatty acid autoxidation alters both the quality and safety of these products. Efficient and fast methods are needed to track lipid oxidation in complex systems. In this study, an oil-in-water emulsion (20% v/v of fish oil stabilized with high oleic sunflower lecithin) was subjected to iron-initiated oxidation. Conjugated dienes (CDs) were measured after fat extraction using a standardized method. Near infrared spectroscopy (NIRS) has been used to record chemical changes occurring during oxidation directly in the emulsion. Variations were noticed in different spectral regions. Partial least squares regression (PLSR) revealed correlations between conjugated diene values and NIRS spectra. High coefficients of determination (R2 = 0.967 and 0.996) were found for calibration and prediction respectively. The CD value was predicted from NIRS spectra with an error of 7.26 mmol eq. LH kg-1 oil (7.8% error). Limits of detection (LOD) and quantification (LOQ) of 4.65 and 15.5 mmol eq. LH kg-1 oil were estimated. NIRS is a rapid and simple method for measuring lipid oxidation (CD value) in an emulsion without prior fat extraction. NIRS can replace the reference methods that use hazardous solvents and consume time. Therefore, NIRS enables in-line monitoring for process and quality control.

A Micromachined Metal Oxide Composite Dual Gas Sensor System for Principal Component Analysis-Based Multi-Monitoring of Noxious Gas Mixtures

Microelectronic gas-sensor devices were developed for the detection of carbon monoxide (CO), nitrogen dioxides (NO₂), ammonia (NH₃) and formaldehyde (HCHO), and their gas-sensing characteristics in six different binary gas systems were examined using pattern-recognition methods. Four nanosized gas-sensing materials for these target gases, i.e., Pd-SnO₂ for CO, In₂O₃ for NO_X, Ru-WO₃ for NH₃, and SnO₂-ZnO for HCHO, were synthesized using a sol-gel method, and sensor devices were fabricated using a microsensor platform. Principal component analysis of the experimental data from the microelectromechanical systems gas-sensor arrays under exposure to single gases and their mixtures indicated that identification of each individual gas in the mixture was successful. Additionally, the gas-sensing behavior toward the mixed gas indicated that the traditional adsorption and desorption mechanism of the n-type metal oxide semiconductor (MOS) governs the sensing mechanism of the mixed gas systems.

Application of an Array of Metal-Oxide Semiconductor Gas Sensors in an Assistant Personal Robot for Early Gas Leak Detection

This paper proposes the application of a low-cost gas sensor array in an assistant personal robot (APR) in order to extend the capabilities of the mobile robot as an early gas leak detector for safety purposes. The gas sensor array is composed of 16 low-cost metal-oxide (MOX) gas sensors, which are continuously in operation. The mobile robot was modified to keep the gas sensor array always switched on, even in the case of battery recharge. The gas sensor array provides 16 individual gas measurements and one output that is a cumulative summary of all measurements, used as an overall indicator of a gas concentration change. The results of preliminary experiments were used to train a partial least squares discriminant analysis (PLS-DA) classifier with air, ethanol, and acetone as output classes. Then, the mobile robot gas leak detection capabilities were experimentally evaluated in a public facility, by forcing the evaporation of (1) ethanol, (2) acetone, and (3) ethanol and acetone at different locations. The positive results obtained in different operation conditions over the course of one month confirmed the early detection capabilities of the proposed mobile system. For example, the APR was able to detect a gas leak produced inside a closed room from the external corridor due to small leakages under the door induced by the forced ventilation system of the building.

A Simple Procedure to Assess Limit of Detection for Multisensor Systems

Currently, there are no established procedures for limit of detection (LOD) evaluation in multisensor system studies, which complicates their correct comparison with other analytical techniques and hinders further development of the method. In this study we propose a simple and visually comprehensible approach for LOD estimation in multisensor analysis. The suggested approach is based on the assessment of evolution of mean relative error values in calibration series with growing analyte concentration. The LOD value is estimated as the concentration starting from which MRE values become stable from sample to sample. This intuitive procedure was successfully tested with a variety of real data from potentiometric multisensor systems.

Real-Time Thermal Modulation of High Bandwidth MOX Gas Sensors for Mobile Robot Applications

A new signal processing technique has been developed for resistive metal oxide (MOX) gas sensors to enable high-bandwidth measurements and enhanced selectivity at PPM levels (<5 PPM VOCs). An embedded micro-heater is thermally pulsed from a temperature of 225 to 350 °C, which enables the chemical reaction kinetics of the sensing film to be extracted using a fast Fourier transform. Signal processing is performed in real-time using a low-cost microcontroller integrated into a sensor module. Three sensors, coated with SnO₂, WO₃ and NiO respectively, were operated and processed at the same time. This approach enables the removal of long-term baseline drift and is more resilient to changes in ambient temperature. It also greatly reduced the measurement time from ~10 s to 2 s or less. Bench-top experimental results are presented for 0 to 200 ppm of acetone, and 0 ppm to 500 ppm of ethanol. Our results demonstrate our sensor system can be used on a mobile robot for real-time gas sensing.

Smelling Nano Aerial Vehicle for Gas Source Localization and Mapping

This paper describes the development and validation of the currently smallest aerial platform with olfaction capabilities. The developed Smelling Nano Aerial Vehicle (SNAV) is based on a lightweight commercial nano-quadcopter (27 g) equipped with a custom gas sensing board that can host up to two in situ metal oxide semiconductor (MOX) gas sensors. Due to its small form-factor, the SNAV is not a hazard for humans, enabling its use in public areas or inside buildings. It can autonomously carry out gas sensing missions of hazardous environments inaccessible to terrestrial robots and bigger drones, for example searching for victims and hazardous gas leaks inside pockets that form within the wreckage of collapsed buildings in the aftermath of an earthquake or explosion. The first contribution of this work is assessing the impact of the nano-propellers on the MOX sensor signals at different distances to a gas source. A second contribution is adapting the ‘bout’ detection algorithm, proposed by Schmuker et al. (2016) to extract specific features from the derivative of the MOX sensor response, for real-time operation. The third and main contribution is the experimental validation of the SNAV for gas source localization (GSL) and mapping in a large indoor environment (160 m²) with a gas source placed in challenging positions for the drone, for example hidden in the ceiling of the room or inside a power outlet box. Two GSL strategies are compared, one based on the instantaneous gas sensor response and the other one based on the bout frequency. From the measurements collected (in motion) along a predefined sweeping path we built (in less than 3 min) a 3D map of the gas distribution and identified the most likely source location. Using the bout frequency yielded on average a higher localization accuracy than using the instantaneous gas sensor response (1.38 m versus 2.05 m error), however accurate tuning of an additional parameter (the noise threshold) is required in the former case. The main conclusion of this paper is that a nano-drone has the potential to perform gas sensing tasks in complex environments.

Influence of Landscape Structures on Water Quality at Multiple Temporal and Spatial Scales: A Case Study of Wujiang River Watershed in Guizhou

Water quality is highly influenced by the composition and configuration of landscape structure, and regulated by various spatiotemporal factors. Using the Wujiang river watershed as a case study, this research assesses the influence of landscape metrics—including composition and spatial configuration—on river water quality. An understanding of the relationship between landscape metrics and water quality can be used to improve water contamination predictability and provide restoration and management strategies. For this study, eight water quality variables were collected from 32 sampling sites from 2014 through 2017. Water quality variables included nutrient pollutant indicators ammoniacal nitrogen (NH3-N), nitrogen (NO3−), and total phosphate (TP), as well as oxygen-consuming organic matter indicators COD (chemical oxygen demand), biochemical oxygen demand (BOD5), dissolved oxygen (DO), and potassium permanganate index (CODMn). Partial least squares (PLS) regression was used to quantitatively analyze the influence of landscape metrics on water quality at five buffer zone scales (extending 3, 6, 9, 12, and 15 km from the sample site) in the Wujiang river watershed. Results revealed that water quality is affected by landscape composition, landscape configuration, and precipitation. During the dry season, landscape metrics at both landscape and class levels predicted organic matter at the five buffer zone scales. During the wet season, only class-level landscape metrics predicted water contaminants, including organic matter and nutrients, at the middle three of five buffer scales. We identified the following important indicators of water quality degradation: percent of landscape, edge density, and aggregation index for built-up land; aggregation index for water; CONTAGION; COHESION; and landscape shape index. These results suggest that pollution can be mitigated by reducing natural landscape composition fragmentation, increasing the connectedness of region rivers, and minimizing human disturbance of landscape structures in the watershed area.

A Practical Method to Estimate the Resolving Power of a Chemical Sensor Array: Application to Feature Selection

A methodology to calculate analytical figures of merit is not well established for detection systems that are based on sensor arrays with low sensor selectivity. In this work, we present a practical approach to estimate the Resolving Power of a sensory system, considering non-linear sensors and heteroscedastic sensor noise. We use the definition introduced by Shannon in the field of communication theory to quantify the number of symbols in a noisy environment, and its version adapted by Gardner and Barlett for chemical sensor systems. Our method combines dimensionality reduction and the use of algorithms to compute the convex hull of the empirical data to estimate the data volume in the sensor response space. We validate our methodology with synthetic data and with actual data captured with temperature-modulated MOX gas sensors. Unlike other methodologies, our method does not require the intrinsic dimensionality of the sensor response to be smaller than the dimensionality of the input space. Moreover, our method circumvents the problem to obtain the sensitivity matrix, which usually is not known. Hence, our method is able to successfully compute the Resolving Power of actual chemical sensor arrays. We provide a relevant figure of merit, and a methodology to calculate it, that was missing in the literature to benchmark broad-response gas sensor arrays.