Spectral Angle Mapper Classification Research Articles

Hyperspectral imaging with machine learning for in vivo skin carcinoma margin assessment: a preliminary study.

Surgical excision is the most effective treatment of skin carcinomas (basal cell carcinoma or squamous cell carcinoma). Preoperative assessment of tumoral margins plays a decisive role for a successful result. The aim of this work was to evaluate the possibility that hyperspectral imaging could become a valuable tool in solving this problem. Hyperspectral images of 11 histologically diagnosed carcinomas (six basal cell carcinomas and five squamous cell carcinomas) were acquired prior clinical evaluation and surgical excision. The hyperspectral data were then analyzed using a newly developed method for delineating skin cancer tumor margins. This proposed method is based on a segmentation process of the hyperspectral images into regions with similar spectral and spatial features, followed by a machine learning-based data classification process resulting in the generation of classification maps illustrating tumor margins. The Spectral Angle Mapper classifier was used in the data classification process using approximately 37% of the segments as the training sample, the rest being used for testing. The receiver operating characteristic was used as the method for evaluating the performance of the proposed method and the area under the curve as a metric. The results revealed that the performance of the method was very good, with median AUC values of 0.8014 for SCCs, 0.8924 for BCCs, and 0.8930 for normal skin. With AUC values above 0.89 for all types of tissue, the method was considered to have performed very well. In conclusion, hyperspectral imaging can become an objective aid in the preoperative evaluation of carcinoma margins.

Mapping and discrimination of the mineralization potential in granitoids from Banyo area (Adamawa, Cameroon), using Landsat 9 OLI, ASTER images and field observations

The Banyo area is situated in the west Cameroon domain of the Central African Fold Belt. In this study, field data and remote sensing conventional mapping techniques including Red-Green-Blue (RGB) colour combination, principal component analysis (PCA), Crosta-PCA, Band Ratios (BR), Constrained Energy Minimization (CEM), Minimum Noise Fraction (MNF), Spectral Angle Mapper (SAM) classification and Linear Spectral Unmixing (LSU) algorithms were used to process Landsat 9 OLI and ASTER data in order to discriminate various rock units such as biotite-amphibole granites, hornblende-biotite orthogneiss and map associated mineral deposits (e.g., iron-oxide, hydroxyl and corundum minerals). In addition, it was found that RGB combination (743) of OLI bands, PCA of OLI bands (PC123), BR of OLI bands (4/2, 6/7, 6/5), BR of ASTER (4/2, 7/6, 8/9), Crosta-PCA of OLI bands, MNF (123) of OLI bands, SAM, and LSU techniques applied to ASTER bands ascertain potential locations of granitoids and gneisses bearing iron and hydroxyl minerals. On the other hand, this study also revealed that corundum deposits are detected by applying the CEM method to OLI band ratio 6/7 highlighting red tones and processing SAM classification algorithm both using laboratory reference spectra. Automatic (using PCI Geomatica line tool) and manual extractions of lineaments revealed three major trends (N-S, E-W, and WNW-ESE to NW-SE) and an accessory trend (NE-SW to ENE-WSW) that surround hydrothermally altered rocks recognized as containing iron oxide, hydroxyl, and corundum mineral deposits. In the field, these lineaments may group to the tree deformation phases. The D1 phase (NW-SE); the D2 phase (NE-SW to ENE-WSW) and the D3 phase (N-S). Maximum Likelihood Calssification (MLC) and SAM classification methods showed to be more effective in discriminating rocks bearing Iron-oxide, hydroxyl and corundum minerals in the study area.This result permits to confirm that, the combination of remote sensing (Landsat 9 OLI and ASTER data in this case) and field data constitute a useful tool for mineral exploration in the shear zone. However, the accuracy assessment was relied on comparing the different maps produced from Landsat 9 OLI and ASTER data with the ground truth sample points and also, with the earliest geological map of Banyo area. Thus, quantitatively the results were accurate at 99%.

Characterization of Rice-Magnaporthe oryzae Interactions by Hyperspectral Imaging.

Hyperspectral imaging has the potential to detect, characterize, and quantify plant diseases objectively and nondestructively to improve phenotyping in breeding for disease resistance. In this study, leaf spectral reflectance characteristics of five rice genotypes diseased with blast caused by three Magnaporthe oryzae isolates differing in virulence were compared with visual disease ratings under greenhouse conditions. Spectral information (140 wavebands, range 450 to 850 nm) of infected leaves was recorded with a hyperspectral imaging microscope at 3, 5, and 7 days postinoculation to examine differences in symptom phenotypes and to characterize the compatibility of host-pathogen interactions. Depending on the rice genotype × M. oryzae genotype interaction, blast symptoms varied from tiny necrosis to enlarged lesions with symptom subareas differing in tissue coloration and indicated gene-for-gene-specific interactions. The blast symptom types were differentiated based on their spectral characteristics in the visible/near-infrared range. Symptom-specific spectral signatures and differences in the composition of leaf blast symptom type(s) resulted in unique spectral and spatial patterns of the rice × M. oryzae interactions based on the size, shape, and color of the symptom subareas. Spectral angle mapper classification of spectra enabled (i) discrimination between healthy (green) and diseased tissue of rice genotypes, (ii) classification and quantification of different blast symptom subareas, and (iii) grading of the host-pathogen compatibility (low - intermediate - high). Hyperspectral imaging was more sensitive to small changes in disease resistance than visual disease assessments and enabled the characterization of various types of resistance/susceptibility reactions of tissue subjected to M. oryzae infection.

Exploring Effective Detection and Spatial Pattern of Prickly Pear Cactus (Opuntia Genus) from Airborne Imagery before and after Prescribed Fires in the Edwards Plateau

Over the past century, prickly pear (PP) cactus (e.g., genus Opuntia; subgenus Platyopuntia) has increased on semi-arid rangelands. Effective detection of cacti abundance and spatial pattern is challenging due to the inherent heterogeneity of rangeland landscapes. In this study, high-resolution multispectral imageries (0.21 m) were used to test object-based (OB) feature extraction, random forest (RF) machine learning, and spectral endmember (n-D) classification methods to map PP and evaluate its spatial pattern. We trained and tested classification methods using field-collected GPS location, plant cover, and spectrometry from 288 2 m radius polygons before a prescribed burn and 480 samples after the burn within a 69.2-ha burn unit. The most accurate classification method was then used to map PP distribution and quantify abundance before and after fire. As a case study, we assessed the spatial pattern of mapped PP cover, considering topoedaphic setting and burn conditions. The results showed that the endmember classification method, spectral angle mapper (SAM), outperformed the RF and OB classifications with higher kappa coefficients (KC) (0.93 vs. 0.82 and 0.23, respectively) and overall accuracies (OA) (0.96 vs. 0.91 and 0.49) from pre-fire imagery. KC and OA metrics of post-fire imagery were lower, but rankings among classification methods were similar. SAM classifications revealed that fire reduced PP abundance by 46.5%, but reductions varied by soil type, with deeper soils having greater decreases (61%). Kolmogorov-Smirnov tests indicated significant changes before and after fire in the frequency distribution of PP cover within deeper soils (D = 0.64, p = 0.02). A two-way ANOVA revealed that the interaction of season (pre- vs. post-fire) and soils significantly (p < 0.00001) influenced the spatial pattern of PP patches. Fire also reduced the size and shape of PP patches depending on the topoedaphic settings. This study provides an innovative and effective approach for integrating field data collection, remote sensing, and endmember classification methods to map prickly pear and assess the effects of prescribed fire on prickly pear spatial patterns. Accurate mapping of PP can aid in the design and implementation of spatially explicit rangeland management strategies, such as fire, that can help reduce and mitigate the ecological and economic impacts of prickly pear expansion.

Mapping Abandoned Uranium Mine Features Using Worldview-3 Imagery in Portions of Karnes, Atascosa and Live Oak Counties, Texas

Worldview-3 (WV3) 16-band multispectral data were used to map exposed bedrock and mine waste piles associated with legacy open-pit mining of sandstone-hosted roll-front uranium deposits along the South Texas Coastal Plain. We used the “spectral hourglass” approach to extract spectral endmembers representative of these features from the image. This approach first requires calibrating the imagery to reflectance, then masking for vegetation, followed by spatial and spectral data reduction using a principal component analysis-based procedure that reduces noise and identifies homogeneous targets which are “pure” enough to be considered spectral endmembers. In this case, we used a single WV3 image which covered an ~11.5 km by ~19.5 km area of Karnes, Atascosa and Live Oak Counties, underlain by mined rocks from the Jackson Group and Catahoula Formation. Up to 58 spectral endmembers were identified using a further multi-dimensional class segregation method and were used as inputs for spectral angle mapper (SAM) classification. SAM classification resulted in the identification of at least 117 mine- and mine waste-related features, most of which were previously unknown. Class similarity was further evaluated, and the dominant minerals in each class were identified by comparison to spectral libraries and measured samples of actual Jackson Group uranium host rocks. Redundant classes were eliminated, and SAM was run a second time using a reduced set of 23 endmembers, which were found to map these same features as effectively as using the full 58 set of endmembers, but with significantly reduced noise and spectral outliers. Our classification results were validated by evaluating detailed scale mapping of three known mine sites (Esse-Spoonamore, Wright-McCrady and Garbysch-Thane) with published ground truth information about the vegetation cover, extent of erosion and exposure of waste pile materials and/or geologic information about host lithology and mineralization. Despite successful demonstration of the utility of WV3 data for inventorying mine features, additional landscape features such as bare agricultural fields and oil and gas drill pads were also identified. The elimination of such features will require combining the spectral classification maps presented in this study with high-quality topographic data. Also, the spectral endmembers identified during the course of this study could be useful for larger-scale mapping efforts using additional well-calibrated WV3 images beyond the coverage of our initial study area.

Monitoring biological degradation of historical stone using hyperspectral imaging

ABSTRACT Stone is one of the most common materials used as a building material in central Europe for centuries. Historical objects are endangered by degradation procedures coming from physical, chemical and biological weathering agents.The weathering process itself should be analysed in detail in order to prevent historical objects by application of proper restoration cleaning techniques. Within our research, a historical sandstone block was analysed during time to monitor biological changes on the surface. The object of interest is situated in the immediate vicinity of Charles Bridge in Prague, which is protected as a UNESCO heritage site. This site was chosen due to high overall humidity all year long. For investigation of the year- round process of biodegradation hyperspectral sensor was used. In the first place, data were processed using four vegetation indices (NDVI, RGRI, CRI1 and VREI1). All vegetation indices indicate that vegetation is thriving and subject to normal seasonal change. The second chosen method of data processing is to use spectral reflectance curves and their subsequence processing by Spectral Angle Mapper (SAM) classification algorithms. A decline in vegetation with the onset of autumn and during the winter months was detected.

Comparative analysis of denoising techniques in burn depth discrimination from burn hyperspectral images.

This study analyzes and compares the performance of five denoising techniques (Lee filter, gamma filter, principal component analysis, maximum noise fraction, and wavelet transform) in order to identify the most appropriate one that lead to the most accurate classification of burned tissue in hyperspectral images. Fifteen hyperspectral images of burned patients were acquired and denoising techniques were applied to each image. Spectral angle mapper classifier was used for data classification and the confusion matrix was used for quantitative evaluation of the performances of the denoising methods. The results revealed that gamma filter performed better than other denoising techniques with values of overall accuracy and kappa coefficient of 91.18% and 0.8958 respectively. The lowest performance was detected for principal component analysis. In conclusion, the gamma filter could be considered an optimal choice for noise reduction in burn hyperspectral images and could be used for a more accurate diagnosis of burn depth.

Individual tree-based forest species diversity estimation by classification and clustering methods using UAV data

Monitoring forest species diversity is essential for biodiversity conservation and ecological management. Currently, unmanned aerial vehicle (UAV) remote sensing technology has been increasingly used in biodiversity monitoring due to its flexibility and low cost. In this study, we compared two methods for estimating forest species diversity indices, namely the spectral angle mapper (SAM) classification approach based on the established species-spectral library, and the self-adaptive Fuzzy C-Means (FCM) clustering algorithm by selected biochemical and structural features. We conducted this study in two complex subtropical forest areas, Mazongling (MZL) and Gonggashan (GGS) National Nature Forest Reserves using UAV-borne hyperspectral and LiDAR data. The results showed that the classification method performed better with higher values of R2 than the clustering algorithm for predicting both species richness (0.62 > 0.46 for MZL and 0.55 > 0.46 for GGS) and Shannon-Wiener index (0.64 > 0.58 for MZL, 0.52 > 0.47 for GGS). However, the Simpson index estimated by the classification method correlated less with the field measurements than the clustering algorithm (R2 = 0.44 and 0.83 for MZL and R2 = 0.44 and 0.62 for GGS). Our study demonstrated that the classification method could provide more accurate monitoring of forest diversity indices but requires spectral information of all dominant tree species at individual canopy scale. By comparison, the clustering method might introduce uncertainties due to the amounts of biochemical and structural inputs derived from the hyperspectral and LiDAR data, but it could acquire forest diversity patterns rapidly without distinguishing the specific tree species. Our findings underlined the advantages of UAV remote sensing for monitoring the species diversity in complex forest ecosystems and discussed the applicability of classification and clustering methods for estimating different individual tree-based species diversity indices.

Monitoring sedimentary areas from mine waste products with Sentinel‐2 satellite images: A case study in theSEof Spain

AbstractTorrential rainfall regimes, among others, are the origin of accelerated soil erosion. The Spanish southeastern Mediterranean region is characterised by an arid climate regime affected by extreme erosion episodes with an important loss of sediments. Soil erosion effects are even more noticeable in areas where soil has been anthropically degraded, as in the mining district ofSierra Minera de Cartagena. The present research focuses on monitoring and mapping the changes in the sediment distribution of iron oxides and hydroxyl (OH−) bearing minerals caused by a cold drop known in Spanish as 'DANA' in September 2019 on theRambla del Beal. This shortramblais fed by sediments from its drainage basin and by mining residues, irrigating a wide agricultural area. When discharges overflow theramblachannel, residues spread over its floodplain and reach the ecological protected coastal lagoonMar Menor. The objective of the study was mapping the mineral distribution of the mining materials eroded from the source areas and sedimented inRambla del Bealduring a DANA. The study was carried out using a pre‐ and a post‐DANA image from the Sentinel‐2 satellite. After masking vegetation, urban areas and water bodies, different band ratios (B4/B3, B11/B12, B8A/B6) and a Spectral Angle Mapper (SAM) classification were applied. Sediment deposits were identified in wider areas after the DANA. Iron oxides increased their extension by 11.08% in the central area (B3/B4 with R2of 0.84) and hydroxyl‐bearing minerals increased by 8.95% in theRambla del Beal's headwaters (B11/B12 withR2of 0.71). The SAM classification (with a 0.1 rad threshold and an overall accuracy of 87.33%) allowed the differentiation and classification of two ferric iron oxides (haematite and goethite) and one iron hydrous sulphate mineral (jarosite). Additionally, band ratio images were spatially overlaid with the soil land uses map layer of the cadastre in order to plot the land uses most affected by the transported sediments during the DANA. These results highlighted agricultural land as the areas (land uses) most affected by iron oxides deposition, as oxidation processes occur more rapidly in these areas. However, grassland and scrubland were the areas with the highest content of hydroxyl‐bearing minerals, as water is accumulated in these places, which favours hydrolysis reactions.HighlightsTorrential events can relocate large volumes of contaminated sediments from former mining areas.Sentinel‐2 allows monitoring mine waste iron‐sulphate changes caused by weathering alteration.Spatial mineralogical pattern in sediments is controlled by geomorphology and flood dynamics.The proposed digital image analysis allows mapping short‐term evolution of mine waste sediments.

Application of hyperspectral and multispectral datasets for mineral mapping

In this study, hyperspectral datasets are simulated from multispectral data using a spectral reconstruction approach which is a sensor-independent technique. This technique makes use of information from atmospherically corrected multispectral Remote Sensing (MRS) data and normalized ground spectra for the simulation of HRS data. In this study EO-1, the ALI dataset was used for the simulation of hyperspectral Remote Sensing (HRS) data to discover the Udaipur region’s unique minerals. A total of 61 spectral bands with 10 nm bandwidth were simulated. The simulated HRS data were validated using visual interpretation, statistical and classification approaches. Simulated HRS data from EO-1 Advanced Land Imager (ALI) has shown a high correlation with EO-1 Hyperion data. Spectral Angle Mapper (SAM) classification was also performed on simulated hyperspectral data for mineral mapping. It was observed that simulated hyperspectral data have shown comparable results with Hyperion and are better than their corresponding multispectral datasets.

Remote Sensing Data for Geological Mapping in the Saka Region in Northeast Morocco: An Integrated Approach

Together with geological survey data, satellite imagery provides useful information for geological mapping. In this context, the aim of this study is to map geological units of the Saka region, situated in the northeast part of Morocco based on Landsat Oli-8 and ASTER images. Specifically, this study aims to: (1) map the lithological facies of the Saka volcanic zone, (2) discriminate the different minerals using Landsat Oli-8 and ASTER imagery, and (3) validate the results with field observations and geological maps. To do so, in this study we used different techniques to achieve the above objectives including color composition (CC), band ratio (BR), minimum noise fraction (MNF), principal component analysis (PCA), and spectral angle mapper (SAM) classification. The results obtained show good discrimination between the different lithological facies, which is confirmed by the supervised classification of the images and validated by field missions and the geological map with a scale of 1/500,000. The classification results show that the study area is dominated by Basaltic rocks, followed by Trachy andesites then Hawaites. These rocks are encased by quaternary sedimentary rocks and an abundance of Quartz, Feldspar, Pyroxene, and Amphibole minerals.

Toward Robust River Plastic Detection: Combining Lab and Field‐Based Hyperspectral Imagery

AbstractPlastic pollution in aquatic ecosystems has increased dramatically in the last five decades, with strong impacts on human and aquatic life. Recent studies endorse the need for innovative approaches to monitor the presence, abundance, and types of plastic in these ecosystems. One approach gaining rapid traction is the use of multi‐ and hyperspectral cameras. However, most experiments using this approach were in controlled environments, making findings challenging to apply in natural environments. We present a method linking lab‐ and field‐based identification of macroplastics using hyperspectral data (1,150–1,675 nm). Experiments using riverbank‐harvested macroplastics were set up in a laboratory environment, and on the banks of the Rhine River. Representative pixel selections of eleven lab‐based images (n = 786,264 pixels) and two field‐based images (n = 40,289 pixels) were used to analyze the differences between these environments. Next, classifier algorithms such as support vector machines (SVM), spectral angle mapper (SAM) and spectral information divergence (SID) were applied, because of their robustness to varying light conditions and high accuracies in mapping spectral similarities. Our results showed that SAM classifiers are most robust in separating plastic pixels from background elements. By applying lab‐based data for plastic detection in field‐based images, user accuracies for plastics to up to 93.6% (n = 8,370 plastic pixels) were attained. This study provides key fundamental insights in linking lab‐based data to plastic detection in the field. With this paper we aim to contribute to the development of future spectral missions to detect and monitor plastic pollution in aquatic ecosystems.

The emergence of artisanal gold mining and local perceptions in the Hounde municipality, Burkina Faso

Artisanal gold mining (AGM) plays an important role in the economic development of Burkina Faso with over 2 million people directly or indirectly engaged in the activity in over 800 sites. This study explores the emergence of AGM and its environmental, social, and economic impacts in the Hounde district of Burkina Faso. A supervised spectral angle mapper classification algorithm in QGIS along with visual interpretation was used to classify images into Land Use Land Cover (LULC) categories to identify the environmental impacts of AGM. The socio-economic impacts of AGM were assessed using semi-structured questionnaires and interviews to identify the perceived impacts of the emergence of AGM in the study area. The results revealed a massive decrease in forest and sparse area of about 18% and 32% respectively from 1990 to 2020, with AGM emerging between 2010 and 2020. LULC analysis also reveals bare land and settlements increased drastically with the emergence of AGM. In terms of social impacts, AGM has no statistically significant impact on land conflicts and cost of living (p-value > 0.05) but a significant impact on current employment status (p-value = 0.00) and social vices like alcoholism (p-value < 0.02) and current school dropouts. The economic impact of AGM on local businesses is also statistically significant (p-value = 0.006) but the chi-square reveals that over 50% of respondents earn less than 1 dollar per day. So, while AGM has proved to be helping to alleviate poverty (SDG1), it compromises the quality of education (SDG4) and the quality of life on land (SDG15) through vegetation reduction. We, therefore, recommend that the policymakers in Burkina Faso oversee mining in the Houde district to adhere to environmental protection guidelines while using revenue generated to improve the quality of life and education in line with SDG15, SDG1, and SDG4 respectively.

Mangrove monitoring in Lothian Island using airborne hyperspectral AVIRIS-NG data

Constant monitoring of the mangroves is important for understanding their health species assemblage and undertaking policies to rehabilitate or restore them. In this study, an attempt was made to classify and detect health and carbon stock in mangroves through a Remote Sensing perspective. The AVIRIS-NG dataset with high spectral (5 nm) and spatial (5 m) resolution proved to be capable of identifying different species of the same mangrove genus along with other morphological features. Using Spectral Angle Mapper (SAM) classification technique, 75% accuracy has been achieved over 12 separate classes. The health of the Lothian Island mangroves was evaluated by using eight different vegetation indices, namely ACI, ARI, MARI, ARVI, CAI, CARI, Chl-red-edge, and REPI. The vegetation indices were normalized and then used as input in a Multi-Criteria Decision Support System (MCDSS) model. The MCDSS model output health map clearly identifies the degraded and healthy mangroves on Lothian Island. The Lothian Island mangroves were found to be degraded in most of the region, which should be of major concern. Above Ground Biomass (AGB), which is another indicator of health, was calculated using Field measured data and AVIRIS-NG hyperspectral data. The highest, lowest, and mean biomass recorded in field transact is 507.19 and 7.39 and 126.43 tonnes, respectively. These three parameters, namely species assemblage of mangroves, their health, and AGB combined can provide much insight into the ecosystem health of Lothian Island.

URBAN GROWTH MONITORING – REMOTE SENSING METHODS FOR SUSTAINABLE DEVELOPMENT

Abstract. Urban areas account for a small fraction of the Earth's surface but have a disproportionate impact on its surroundings regarding mass, energy, and resources. An exponential increase in the urban population has been observed since the mid-20th century. As expected by the United Nations (UN), by the year 2050, 68.4% of the world population will live in cities with a population of 20,000 or more. Due to enormous socio-economic pressures resulting from population expansion, urbanization and intensive changes in the landscape, an urban development program to make cities and human settlements inclusive, safe, resilient, and sustainable has become of the utmost importance since the 2005 World Summit in Rio and further adopted in 2015 by UN as the "2030 Agenda" for Sustainable Development. The study focused on monitoring the SDG 11 target 11.3.1. they are defined as a ratio of land consumption rate to the population growth rate because mapping urban land quickly and accurately is indispensable for watershed run-off prediction and other planning applications. There is no well-established, consistent way to measure either urban land sprawl or population growth. However, remote sensing methods and satellite-derived data make it possible to monitor urban growth rates over large areas in a relatively short time. There are many techniques for urban land cover automatically mapping. These techniques can be broadly grouped into two general types: those based on the input data classification, including pixel- and object-based classifications and those based on directly segmenting the indices, such as the commonly used normalized difference vegetation index (NDVI), normalized build-up area index (NDBI), and their modifications.The authors used classical pixel (supervised classification with Spectral Angle Mapper classification and KNN methods) and objectbased classification in the presented research. In addition, spectral indices, i.e., NDVI, NDBI and their modifications to derive buildup areas, were applied. Moreover, the authors focused on the recent deep learning and machine learning methods, i.e., the utilization of spatial-context information in multi-temporal data to learn hierarchical feature representations. All methods of detecting built-up areas were compared and assessed based on the available cartographic data.

Harmonizing Definitions and Methods to Estimate Deforestation at the Lacandona Tropical Region in Southern Mexico

Deforestation is a major factor reducing natural habitats, leading to tropical ecosystems and biodiversity loss worldwide. The Lacandona region in southern Mexico holds one of the largest fragments of tropical rainforest in North America. We evaluated the deforestation of the Lacandona region harmonizing concepts and methodologies. An international (FAO definition), governmental (national definition), and regional definition of deforestation with applications at different scales were analyzed and harmonized with two classification methods (likelihood and spectral angle mapper (SAM)). We used 2015 and 2018 Landsat 8 images, and likelihood and SAM classifications were applied for FAO and regional definitions of deforestation. Overall, the best evaluated classifier in quantity was likelihood for 2015 and 2018 (kappa: 0.87 and 0.70, overall accuracy: 91.8 and 80.4%, and quantity disagreement: 4.1 and 10 %, respectively). The allocation disagreement only showed exchange between classes. Nevertheless, they did not show differences between classifiers, although 2015 had less disagreement than 2018: exchange, 4.1% for likelihood and SAM; shift: 0% for likelihood and SAM. Maps based on the regional definition of deforestation showed that the likelihood classification detected 11,441 ha less deforestation than SAM (40,538 and 51,979 ha, respectively). The FAO definition of deforestation showed that likelihood classification detected 11,914 ha less deforestation than SAM classification (37,152 and 49,066 ha, respectively). Further, the likelihood classification showed 3387 ha more of deforestation according to the regional definition than the FAO definition of deforestation (40,538 and 37,152 ha, respectively). SAM classification showed that the regional definition showed 2913 ha more deforestation than the FAO definition (51,979 and 49,066, respectively). We concluded that implementation of governmental programs in the Lacandona region requires estimations based on a careful selection of deforestation definitions and methods.

Can Hyperspectral Imaging and Neural Network Classification Be Used for Ore Grade Discrimination at the Point of Excavation?

This work determines whether hyperspectral imaging is suitable for discriminating ore from waste at the point of excavation. A prototype scanning system was developed for this study. This system combined hyperspectral cameras and a three-dimensional LiDAR, mounted on a pan-tilt head, and a positioning system which determined the spatial location of the resultant hyperspectral data cube. This system was used to obtain scans both in the laboratory and at a gold mine in Western Australia. Samples from this mine site were assayed to determine their gold concentration and were scanned using the hyperspectral apparatus in the laboratory to create a library of labelled reference spectra. This library was used as (i) the reference set for spectral angle mapper classification and (ii) a training set for a convolutional neural network classifier. Both classification approaches were found to classify ore and waste on the scanned face with good accuracy when compared to the mine geological model. Greater resolution on the classification of ore grade quality was compromised by the quality and quantity of training data. The work provides evidence that an excavator-mounted hyperspectral system could be used to guide a human or autonomous excavator operator to selectively dig ore and minimise dilution.

An assessment of image classifiers for generating machine-learning training samples for mapping the invasive Campuloclinium macrocephalum (Less.) DC (pompom weed) using DESIS hyperspectral imagery

Machine-learning algorithms may require large numbers of reference samples to train depending on the spatial and spectral heterogeneity of the mapping area. Acquiring these reference samples using traditional field data collection methods is a challenge due to time constraints, logistical limitations, and terrain inaccessibility. The aim of study was to assess how parametric, nonparametric, and spectral matching image classifiers can be used to generate a large number of accurate training samples from minimal ground control points to train machine-learning algorithms for mapping the invasive pompom weed using 30 m DESIS hyperspectral data. Three image classifiers, namely, maximum likelihood classifier (MLC), support vector machine (SVM) and spectral angle mapper (SAM) were selected to represent each of the three types of image classifiers under investigation in this study. Results show that the SAM, MLC and SVM classifiers had pixel-based classification accuracies of 87%, 73% and 67% for the pompom-containing pixels class, respectively. Furthermore, an independent field verification for the SAM classification was conducted yielding a 92% overall mapping accuracy for the pompom-containing pixels class. A total of 4000 pompom-containing and 8000 non-pompom-containing training samples were generated from an SAM classification that was trained using only 20 endmembers. Overall, this study presents a potential solution strategy that has significant implications for generating large numbers of reference training samples for mapping invasive alien plants from new generation spaceborne hyperspectral imagery using machine-learning algorithms.

Comparative analysis of AVIRIS-NG and Landsat-8 OLI data for lithological mapping in parts of Sittampundi layered complex, Tamil Nadu, India

The present study addresses a comparative analysis of the potentials of Airborne Visible and infrared imaging spectrometer -next generation (AVIRIS-NG) data, with respect to Landsat-8 Operational Land Imager (OLI) data for mapping of sporadic exposures of mafic cumulates within the anorthosite intrusion and delineation of major lithological units of Sittampundi layered complex. AVIRIS-NG and Landsat-8 OLI data were initially compared using scene-based relative signal-to-noise ratio (SNR) analysis. Subsequently, spectral anomaly maps were generated to map different lithological rock units using the constrained energy minimization (CEM) approach. Spectral maps derived using CEM were further filtered by low-pass filtering to control the false-positive result. Spectral angle mapper (SAM) classification technique was applied to analyze the suitability of AVIRIS-NG and Landsat-8 OLI data for lithological mapping. Confusion matrix analysis was carried out to assess the SAM classified maps. The first vertical derivative (FVD) map of magnetic anomaly and field evidence/geological exposures were used to validate the results. SNR analysis indicates the superior signal strength of AVIRIS-NG data than Landsat-8 OLI data. Moreover, combined analysis based on CEM and SAM of remote sensing data and FVD of magnetic anomaly reveals that the false-positive area indicating metagabbro/mafic cumulate is lower in AVIRIS-NG data than Landsat-8 OLI data. This is because of the relatively poor SNR, spectral and spatial resolutions of Landsat-8 OLI data; however, it can be good enough for mapping of metagabbro/mafic cumulate when AVIRIS-NG data is unavailable.

Spectral classification and MGM-based mineralogical characterization of hydrated phases: The Nili Fossae case, Mars

Mars has a relatively tenuous atmosphere, the remote-sensed spectral data of the surface are affected by the presence of the atmosphere. In particular the water and CO2 absorption of the atmosphere are in the same spectral region as water, OH− and (CO3)2- in minerals. An accurate correction of the gaseous contamination is thus mandatory for applying powerful mineralogical assessment methods, such as MGM (Modified Gaussian Model), also in the spectral ranges affected by possible residual from atmospheric features. In this paper, for the first time we apply MGM on hydrated phases using OMEGA/MEX corrected data through the Surface Atmosphere Separation (SAS) method, which allows a more reliable gaseous absorption removal when working on reflectance spectra. Nili Fossae area is selected as a test case to prove the validity of this approach. SAS-corrected OMEGA hyper-spectral images are analyzed using a set of sequential steps, comprising image derived spectral library collection, supervised classification and spectral band assignment. Applying the PPI (Pixel Purity Index) on the MNF (Minimum Noise Fraction) transform, 14 end-members are selected and used to spectrally classify the region through SAM (Spectral Angle Mapper) classification. The MGM algorithm is applied for the mineralogical interpretation of Nili Fossae in the 0.5–2.5 ​μm spectral range, with particular focus on hydrated mineralogies. We recognize 5 main spectral units: A) dominated by orthopyroxene and iron hydroxides, B) and C) characterized by a mixture of orthopyroxene and clinopyroxene (with different mineral proportions for the two units), D) dominated by olivine and altered olivine and E) showing hydrated mineralogies: good matches are found with nontronite, chamosite and serpentine.