Visible To Near-infrared Research Articles

Identification of peat type and humification by laboratory VNIR/SWIR hyperspectral imaging of peat profiles with focus on fen-bog transition in aapa mires

AimsHyperspectral imaging (HSI) has high potential for analysing peat cores, but methodologies are deficient. We aimed for robust peat type classification and humification estimation. We also explored other factors affecting peat spectral properties.MethodsWe used two laboratory setups: VNIR (visible to near-infrared) and SWIR (shortwave infrared) for high resolution imaging of intact peat profiles with fen-bog transitions. Peat types were classified with support vector machines, indices were developed for von Post estimation, and K-means clustering was used to analyse stratigraphic patterns in peat quality. With separate experiments, we studied spectral effects of drying and oxidation.ResultsDespite major effects, oxidation and water content did not impede robust HSI classification. The accuracy between Carex peat and Sphagnum peat in validation was 80% with VNIR and 81% with SWIR data. The spectral humification indices had accuracies of 82% with VNIR and 56%. Stratigraphic HSI patterns revealed that 36% of peat layer shifts were inclined by over 20 degrees. Spectral indices were used to extrapolate visualisations of element concentrations.ConclusionsHSI provided reliable information of basic peat quality and was useful in visual mapping, that can guide sampling for other analyses. HSI can manage large amounts of samples to widen the scope of detailed analysis beyond single profiles and it has wide potential in peat research beyond the exploratory scope of this paper. We were able to confirm the capacity of HSI to reveal shifts of peat quality, connected to ecosystem-scale change.

VNIR-SWIR Superspectral Mineral Mapping: An Example from Cuprite, Nevada

Cuprite, Nevada, is a location well known for numerous studies of its hydrothermal mineralogy. This region has been used to validate geological interpretations of airborne hyperspectral imagery (AVIRIS HSI ), Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER ) imagery, and most recently eight-band WorldView-3 shortwave infrared (SWIR ) imagery. WorldView-3 is a high-spatial-resolution commercial multispectral satellite sensor with eight visible-to-near-infrared (VNIR ) bands (0.42–1.04 μm) and eight SWIR bands (1.2–2.33 μm). We have applied mineral mapping techniques to all 16 bands to perform a geological analysis of the Cuprite, Nevada, location. Ground truth for the training and validation was derived from AVIRIS hyperspectral data and United States Geological Survey mineral spectral data for this location. We present the results of a supervised mineral-mapping classification applying a random-forest classifier. Our results show that with good ground truth, WorldView-3 SWIR + VNIR imagery produces an accurate geological assessment.

Investigating the Potential of a Newly Developed UAV-based VNIR/SWIR Imaging System for Forage Mass Monitoring

Remote sensing systems based on unmanned aerial vehicles (UAVs) are well suited for airborne monitoring of small to medium-sized farmland in agricultural applications. An imaging system is often used in the form of a multispectral multi-camera system to derive well-established vegetation indices (VIs) efficiently. This study investigates the potential of such a multi-camera system with a novel approach to extend spectral sensitivity from visible-to-near-infrared (VNIR) to short-wave infrared (SWIR) (400–1700 nm) for estimating forage mass from an aerial carrier platform. The system test was performed in a grassland fertilizer trial in Germany near Cologne in late July 2019. Within 37 min, a spectral response in four different wavelength bands in the NIR and SWIR range was acquired during two consecutive flights. Spectral image data were calibrated to reflectance using two different methods. The resulting reflectance data sets were processed to orthomosaics for each wavelength band. From these orthomosaics for both calibration methods, the four-band NIR/SWIR GnyLi VI and the two-band NIR/SWIR Normalized Ratio Index (NRI), were calculated. During both UAV flights, spectral ground truth data were recorded with a spectroradiometer on 12 plots in total for validation of camera-based spectral data. The camera and spectroradiometer data sets were directly compared in resulting reflectance and further analyzed with simple linear regression (SLR) models to predict dry matter (DM) yield. In the camera-based SLRs, the NRI performed best with R^2 of 0.73 and 0.75 (RMSE: 0.18 and 0.17) before the GnyLi with R^{2} of 0.71 and 0.73 (RMSE: 0.19 and 0.18). These results clearly indicate the potential of the camera system for applications in forage mass monitoring.

Gate-Tunable Optical Nonlinearities and Extinction in Graphene/LaAlO3/SrTiO3 Nanostructures.

We explore the ultrafast optical response of graphene subjected to intense (∼106 V/cm) local (∼10 nm) electric fields. Nanoscale gating of graphene is achieved using a voltage-biased, SrTiO3-based conductive nanowire junction "written" directly under the graphene and isolated from it by an insulating ultrathin (<2 nm) LaAlO3 barrier. Upon illumination with ultrafast visible-to-near-infrared (VIS-NIR) light pulses, the local field from the nanojunction creates a strong gate-tunable second-order nonlinearity in the graphene and produces a substantial difference-frequency (DFG) and sum-frequency generation (SFG) response detected by the nanojunction. Spectrally sharp, gate-tunable extinction features (>99.9%) are observed in the VIS-NIR and SFG spectral ranges, in parameter regimes that are positively correlated with the enhanced nonlinear response. The observed graphene-light interaction and nonlinear response are of fundamental interest and open the way for future exploitation in graphene-based optical devices such as phase shifters, modulators, and nanoscale THz sources.

Experimental Investigation of Oxidation of Pyroxene and Basalt: Implications for Spectroscopic Analyses of the Surface of Venus and the Ages of Lava Flows

Abstract Alteration of basalt on the surface of Venus should be dominated by nanophase hematite and sulfate coatings, but the timescale of oxidation and how that affects spectroscopic measurements has not been extensively studied. The oxidation rate and the effect on reflectance spectroscopy are needed to constrain the age of lava flows with high emissivity that were previously measured by the European Space Agency's Venus Express Mission. Here we investigate the effects of oxidation on the mineralogy and visible to near-infrared (VNIR) spectroscopy of augite, diopside, pyroxenite, and an alkali basalt. The experimental and spectroscopic data are used to constrain the surface basaltic weathering rates for Venus. Further, we use the spectroscopic measurements to constrain how quickly igneous signatures would degrade during oxidation of the surface. Our results show that basalt and olivine oxidize within weeks to months on the surface of Venus and their VNIR results are dominated by hematite. Pyroxenes take longer to alter with minor hematite forming on the surface—largely in cracks. Instead, Fe3+ is predominantly forming in the crystal structure. Our work suggests that if basalt on the surface of Venus contains olivine and/or glass, the high emissivity lavas previously measured would be only a few years old. If, instead, those high emissivity lavas are fully crystalline and contain no olivine or glass (which is unlikely), they would still only be decades to hundreds of years old. Therefore, this is consistent with recent work suggesting that Venus is volcanically active today.

Present-day volcanism on Venus as evidenced from weathering rates of olivine.

At least some of Venus' lava flows are thought to be <2.5 million years old based on visible to near-infrared (VNIR) emissivity measured by the Venus Express spacecraft. However, the exact ages of these flows are poorly constrained because the rate at which olivine alters at Venus surface conditions, and how that alteration affects VNIR spectra, remains unknown. We obtained VNIR reflectance spectra of natural olivine that was altered and oxidized in the laboratory. We show that olivine becomes coated, within days, with alteration products, primarily hematite (Fe2O3). With increasing alteration, the VNIR 1000-nm absorption, characteristic of olivine, also weakens within days. Our results indicate that lava flows lacking VNIR features due to hematite are no more than several years old. Therefore, Venus is volcanically active now.

Early detection of chilling injury in green bell peppers by hyperspectral imaging and chemometrics

Abstract The feasibility of using hyperspectral imaging in the combined wavelength regions comprising of visible to near infrared (VIS-NIR) (400−1000 nm) and short wave infrared (SWIR) (1000−2500 nm) was investigated for discriminating fresh bell peppers from those stored under refrigeration. In addition, the technique was used for early detection of chilling injury (CI) in mature fruit. Supervised classification models were developed using Partial Least Square Discriminant Analysis (PLS-DA) for raw and pre-processed spectra followed by wavelength selection using VIP scores. Reliable classification of fresh and stored fruit was achieved using pre-processed data in VIS-NIR range by 88 % and 84 % non-error-rate (NER) for calibration (Cal) and cross-validation (CV), respectively, but a slightly higher classification accuracy was manifested in the SWIR range using raw spectra; in this case, wavelength selection resulted in six wavelengths in VIS-NIR reaching to 87 %, 83 NER for Cal, CV, respectively, and four wavelengths in SWIR range yielding to NER of 84 % for Cal and 81 % to satisfy the aforementioned objective. Secondly, classification of fruit based on days of cold storage was achieved using pre-processed data both for VIS-NIR and SWIR whole ranges where wavelength reduction resulted in 12 wavelengths in the VIS-NIR and 13 wavelengths in SWIR range without impressive varying model performance in case of VIS-NIR and decreasing model accuracy in SWIR range. In addition, Partial Least Square Regression (PLSR) was conducted on the data extracted from VIS-NIR HSI, to predict days of cold storage both in full spectral range and selected wavelengths obtained from VIP scores. PLSR models based on full range spectra yielded R²CV = 0.92, while for the PLSR model based on selected wavelengths R²CV = 0.79 was obtained, along with reasonable RMSEC and RMSECV. Conclusively, based on the results, VIS-NIR hyperspectral imaging is a reliable option for on-line classification of fresh versus refrigerated fruit and for identifying early incidence of CI.

Evaluating the performance of hyperspectral short-wave infrared sensors for the pre-sorting of complex ores using machine learning methods

Sensor-based sorting is increasingly used for the concentration of ores. To assess the sorting performance for a specific ore type, the raw materials industry currently conducts trial-and-error batch tests. In this study, a new methodology to assess the potential of hyperspectral visible to near-infrared (VNIR) and short-wave infrared (SWIR) sensors, combined with machine-learning routines to improve the sorting potential evaluation, is presented. The methodology is tested on two complex ores. The first is a tin ore in which cassiterite—the target mineral—is variable in grain size, heterogeneously distributed and has no diagnostic response in the VNIR-SWIR range of the electromagnetic spectrum. However, cassiterite is intimately associated with SWIR active minerals, such as chlorite and fluorite, which can be used as proxies for its presence. The second case study consists of a copper-gold porphyry, where copper occurs mainly in chalcopyrite, bornite, covellite and chalcocite, while gold is present as inclusions in the copper minerals and in pyrite. Machine-learning techniques such as Random Forest and Support Vector Machine applied to the hyperspectral data predict excellent sorting results in terms of grade and recovery. The approach can be adjusted to optimize sorting for a variety of ore types and thus could increase the attractivity of VNIR-SWIR sensor sorting in the minerals industry.

Regional scale prospecting for non-sulphide zinc deposits using ASTER data and different spectral processing methods

ABSTRACT The Ravar-Kuhbanan-Bahabad belt (RKBB) in Central Iran contains several carbonate-hosted non-sulphide Zn (zinc)-Pb (lead) deposits. The Gujer Zn mine area located in the middle of the RKBB was selected as the case study. Due to its large extent, dolomitic envelope in carbonate host rocks can be considered as a more appropriate exploratory target than small Zn-rich gossans or blind karst filling ore. Based on previous studies, the occurrence of red sandstone as a candidate of supplying metal for mineralization and evaporate as sulphate source for mineralized liquids in the vicinity of carbonate rocks can be important exploratory key in the RKBB. Non-sulphide Zn deposits were formed through oxidation of primary Mississippi Valley-type (MVT) deposits in the study area. Remote sensing studies were undertaken using visible to near-infrared (VNIR) and shortwave infrared (SWIR) bands of the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) with the objective of lithological mapping. Five traverse lines containing a total of 81 samples were designed and followed with subsequent chemical analysis, thin section studies, and spectroscopy to verify the results. Two types of carbonates, namely, magnesian dolomite as host rock and surrounding calcitic limestone, were realized through using magnesium oxide (MgO) to calcium oxide (CaO) ratio. Based on spectroscopy studies, calcite and dolomite showed distinct absorption features at 2.35 µm and 2.32 µm, respectively, in ASTER band 8 while a shoulder at 2.25 µm was seen in ASTER band 7 for dolomite. Three image processing methods including spectral angle mapper (SAM), linear spectral unmixing (LSU), and mixture-tuned matched-filtering (MTMF) were applied to separate dolomite and limestone. The accuracy of image classification was numerically estimated using a confusion matrix. Limestone with the accuracy of 95.83% was more precisely enhanced using MTMF method compared to SAM and LSU methods. Highest accuracy of 75% for dolomite was obtained through using LSU method. Red sandstone and evaporate units were classified using MTMF and SAM/LSU methods, respectively. Rock units with the highest accuracy were selected and simply overlain on an image of ASTER in a GIS platform to create the potential map of the study area. Results showed that ASTER data can be successfully used to prepare a potential map for regional scale prospecting for carbonate-hosted non-sulphide Zn-Pb deposits in geological setting and climate condition similar to the RKBB.

Quantification of coffinite (USiO4) in roll-front uranium deposits using visible to near infrared (Vis-NIR) portable field spectroscopy

Coffinite (USiO4) is a common uranium-bearing mineral of roll-front uranium deposits. This mineral can be identified by the visible near infrared (Vis-NIR) portable field spectrometers used in mining exploration. However, due to the low detection limits and associated errors, the quantification of coffinite abundance in the mineralized sandstones or sandy sediments of roll-front uranium deposits using Vis-NIR spectrometry requires a specific methodological development.In this study, the 1135 nm absorption band area is used to quantify the abundance of coffinite. This absorption feature does not interfere with NIR absorption bands of any other minerals present in natural sands or sandstones of uranium roll-front deposits. The correlation between the 1135 nm band area and coffinite content was determined from a series of spectra measured from prepared mineral mixtures. The samples were prepared with a range of weighted amounts of arenitic sands and synthetic coffinite simulating the range of uranium concentration encountered in roll-front uranium deposits.The methodology presented in this study provides the quantification of the coffinite content present in sands between 0.03 wt% to 1 wt% coffinite with a detection limit as low as 0.005 wt%. The integrated area of the 1135 nm band is positively correlated with the coffinite content of the sand in this range, showing that the method is efficient to quantify coffinite concentrations typical of roll-front uranium deposits. The regression equation defined in this study was then used as a reference to predict the amount of natural coffinite in a set of mineralized samples from the Tortkuduk uranium roll-front deposit (South Kazakhstan).

Calibration modelling for non-destructive estimation of external and internal quality parameters of ‘Marsh’ grapefruit using Vis/NIR spectroscopy

Consumer preference for fruit without disorder influences purchase of fruit at both local and international markets. Recent trends in horticulture show that consumer preference is influenced by assurance that external appearance is linked with rewarding internal sensory quality. Therefore, the need for non-destructive evaluation of external and internal quality parameters is important. This study was conducted to develop and test calibration models for integrated prediction of external and internal quality of ‘Marsh’ grapefruit. Visible to near infrared (Vis/NIR) spectroscopy (Vis/NIRS) was used to acquire spectral information from 522 intact fruit. Reference quality parameters such as colour indices (luminosity (L*), greenness (a*) and yellowness (b*)), rind dry matter (DM), rind total phenolics concentration, BrimA, carbohydrates, sweetness index (SI) and total sweetness index (TSI) were obtained using conventional methods. Principal component analysis was applied to analyse spectral data to identify outliers. Savitzky-Golay second derivative with second order polynomial was employed as pre-processing method to correct light scattering properties of the spectra. The spectra were subjected to a test set validation by categorising the spectra into calibration (60%) and validation (40%) sets. Partial least square regression was used as chemometric tool to develop models for predicting each parameter. The model validation results showed that external and internal quality parameters of grapefruit could be predicted with satisfactory accuracy with R2 value of 0.99 for rind quality parameters (L*, a*, b*, DM) and 0.77, 0.99, 0.99 for BrimA, SI and TSI, respectively. The residual predictive deviation (RPD) results for L*, a*, b*, DM, BrimA, SI and TSI were 64.1, 61.4, 123.4, 12.9, 1.4, 9.0 and 13.9, respectively. Vis/NIR calibration and validation results demonstrated that quality parameters of ‘Marsh’ grapefruit could be predicted using Vis/NIRS.

Spectral characterization of analog samples in anticipation of OSIRIS-REx's arrival at Bennu: A blind test study

We present spectral measurements of a suite of mineral mixtures and meteorites that are possible analogs for asteroid (101955) Bennu, the target asteroid for NASA's Origins, Spectral Interpretation, Resource Identification, Security, Regolith Explorer (OSIRIS-REx) mission. The sample suite, which includes anhydrous and hydrated mineral mixtures and a suite of chondritic meteorites (CM, CI, CV, CR, and L5), was chosen to characterize the spectral effects due to varying amounts of aqueous alteration and minor amounts of organic material. Our results demonstrate the utility of mineral mixtures for understanding the mixing behavior of meteoritic materials and identifying spectrally dominant species across the visible to near-infrared (VNIR) and thermal infrared (TIR) spectral ranges. Our measurements demonstrate that, even with subtle signatures in the spectra of chondritic meteorites, we can identify diagnostic features related to the minerals comprising each of the samples. Also, the complementary nature of the two spectral ranges regarding their ability to detect different mixture and meteorite components can be used to characterize analog sample compositions better. However, we observe differences in the VNIR and TIR spectra between the mineral mixtures and the meteorites. These differences likely result from (1) differences in the types and physical disposition of constituents in the mixtures versus in meteorites, (2) missing phases observed in meteorites that we did not add to the mixtures, and (3) albedo differences among the samples. In addition to the initial characterization of the analog samples, we will use these spectral measurements to test phase detection and abundance determination algorithms in anticipation of mapping Bennu's surface properties and selecting a sampling site.

Model development for non-destructive determination of rind biochemical properties of 'Marsh' grapefruit using visible to near-infrared spectroscopy and chemometrics.

Rind biochemical properties play major roles in defence mechanisms against the incidence of rind physiological disorders of citrus fruit during cold storage. Hence, multivariate calibration models were developed to rapidly and non-destructively determine rind biochemical properties of citrus fruit from visible to near-infrared (Vis/NIR) spectra acquired by Vis/NIR spectroscopy using partial least square regression algorithm. To achieve optimum models for determination of each rind biochemical property, several mathematical pre-processing methods were explored, including no pre-treatment. However, special emphases were given to the best model statistics in terms of coefficient of determination (R2) and residual predictive deviation (RPD). Models were performed by critical examination of different wavelength ranges (visible, near-infrared and full regions) and combinations of fruit harvested from different production regions and acquired before (week 0) and after (week 9) cold storage. Results obtained showed excellent models for determining parameters such as sucrose (R2 = 0.99 and RPD = 11.42), total flavonoids (R2 = 0.99 and RPD = 12.37), and chlorophyll b (R2 = 0.97 and RPD = 5.67). This study reported the first application of Vis/NIR and chemometrics in determining the rind biochemical properties of 'Marsh' grapefruit rapidly and non-destructively.

The VIS-SWIR spectrum of skylight polarization.

The skylight degree of linear polarization (DoLP) was previously shown to vary primarily with aerosol optical depth and underlying surface reflectance for visible-to-near-infrared (VNIR) wavelengths. This paper extends the study of skylight polarization to 2.5μm in the short-wave infrared (SWIR). A successive-orders-of-scattering radiative transfer code was used to model skylight polarization with measured inputs that included aerosol properties retrieved from a ground-based solar radiometer (extrapolated into the SWIR) and spectral surface reflectance from a handheld spectrometer. The modeled DoLP depended heavily on the aerosol size distribution at SWIR wavelengths and on the aerosol optical depth at VNIR wavelengths. Once the aerosol optical depth became greater than the Rayleigh optical depth, the predicted polarization deviated significantly from Rayleigh scattering theory. The SWIR polarization spectrum generally decreased at wavelengths beyond 1μm at a rate dependent on the aerosol size distribution. The surface reflectance affected the polarization in the same manner throughout the visible (VIS)-SWIR spectrum, with higher reflectance decreasing the skylight polarization. Validation measurements of SWIR skylight polarization in a 1.5-1.8μm band are also shown. These measurements were made on clean and smoky days using a SWIR imaging polarimeter. In both simulations and measurements, the SWIR skylight polarization was greater in the smoky atmosphere than in the clean atmosphere.

Integration of Terrestrial and Drone-Borne Hyperspectral and Photogrammetric Sensing Methods for Exploration Mapping and Mining Monitoring

Mapping lithology and geological structures accurately remains a challenge in difficult terrain or in active mining areas. We demonstrate that the integration of terrestrial and drone-borne multi-sensor remote sensing techniques significantly improves the reliability, safety, and efficiency of geological activities during exploration and mining monitoring. We describe an integrated workflow to produce a geometrically and spectrally accurate combination of a Structure-from-Motion Multi-View Stereo point cloud and hyperspectral data cubes in the visible to near-infrared (VNIR) and short-wave infrared (SWIR), as well as long-wave infrared (LWIR) ranges acquired by terrestrial and drone-borne imaging sensors. Vertical outcrops in a quarry in the Freiberg mining district, Saxony (Germany), featuring sulfide-rich hydrothermal zones in a granitoid host, are used to showcase the versatility of our approach. The image data are processed using spectroscopic and machine learning algorithms to generate meaningful 2.5D (i.e., surface) maps that are available to geologists on the ground just shortly after data acquisition. We validate the remote sensing data with thin section analysis and laboratory X-ray diffraction, as well as point spectroscopic data. The combination of ground- and drone-based photogrammetric and hyperspectral VNIR, SWIR, and LWIR imaging allows for safer and more efficient ground surveys, as well as a better, statistically sound sampling strategy for further structural, geochemical, and petrological investigations.

Biosignature Preservation Potential in Playa Evaporites: Impacts of Diagenesis and Implications for Mars Exploration.

Assessing biosignature preservation potential (BPP) in ancient habitable environments on Mars is a top NASA priority. We address this goal through the study of Miocene-Pliocene evaporites of the Verde Formation (central Arizona). We assessed the effects of diagenesis on BPP, integrating outcrop-scale observations with six lab analyses: thin-section petrography, X-ray diffraction, Raman spectroscopy, total organic carbon (TOC), electron probe microanalysis (EPMA), and visible to near-infrared (VNIR) reflectance spectroscopy. We recognized five facies and their diagenetic pathways. Two facies included mudstones which contain clusters of displacive growth gypsum (DGG). Early DGG was altered during diagenesis by dissolution forming crystal cavities and later underwent recrystallization, cation substitution, and sulfate dehydration. Another facies was identified by lenticular beds dominated by halite and late diagenetic thenardite (Na2SO4). These pods are overlain by a sequence of interbedded gray and red mudstones which record cyclic oxidation and Fe-oxide cementation. During the Pleistocene, a lacustrine environment developed, accompanied by magnesite cementation of playa mudstones. TOC analyses were used as a proxy for inferring the BPP in each facies. The highest BPP was associated with both red and gray mudstone facies. This study provides a taphonomic framework for playa environments on Earth that record the impacts of diagenesis on BPP, with potential applications to Mars sample return (MSR) missions.

Individual Tree Crown Segmentation and Classification of 13 Tree Species Using Airborne Hyperspectral Data

Knowledge of the distribution of tree species within a forest is key for multiple economic and ecological applications. This information is traditionally acquired through time-consuming and thereby expensive field work. Our study evaluates the suitability of a visible to near-infrared (VNIR) hyperspectral dataset with a spatial resolution of 0.4 m for the classification of 13 tree species (8 broadleaf, 5 coniferous) on an individual tree crown level in the UNESCO Biosphere Reserve ‘Wienerwald’, a temperate Austrian forest. The study also assesses the automation potential for the delineation of tree crowns using a mean shift segmentation algorithm in order to permit model application over large areas. Object-based Random Forest classification was carried out on variables that were derived from 699 manually delineated as well as automatically segmented reference trees. The models were trained separately for two strata: small and/or conifer stands and high broadleaf forests. The two strata were delineated beforehand using CHM-based tree height and NDVI. The predictor variables encompassed spectral reflectance, vegetation indices, textural metrics and principal components. After feature selection, the overall classification accuracy (OA) of the classification based on manual delineations of the 13 tree species was 91.7% (Cohen’s kappa (κ) = 0.909). The highest user’s and producer’s accuracies were most frequently obtained for Weymouth pine and Scots Pine, while European ash was most often associated with the lowest accuracies. The classification that was based on mean shift segmentation yielded similarly good results (OA = 89.4% κ = 0.883). Based on the automatically segmented trees, the Random Forest models were also applied to the whole study site (1050 ha). The resulting tree map of the study area confirmed a high abundance of European beech (58%) with smaller amounts of oak (6%) and Scots pine (5%). We conclude that highly accurate tree species classifications can be obtained from hyperspectral data covering the visible and near-infrared parts of the electromagnetic spectrum. Our results also indicate a high automation potential of the method, as the results from the automatically segmented tree crowns were similar to those that were obtained for the manually delineated tree crowns.

Laboratory-based hyperspectral image analysis for predicting soil carbon, nitrogen and their isotopic compositions

The common methods of determining soil carbon (C), nitrogen (N) and their isotopic compositions (δ13C and δ15N) are expensive and time-consuming. Therefore, alternative low-cost and rapid methods are sought to address this issue. This study aimed to investigate the potential of hyperspectral image analysis to predict soil total carbon (TC), total nitrogen (TN), δ13C and δ15N. Hyperspectral images were captured from 96 ground soil samples using a laboratory-based visible to near-infrared (VNIR) hyperspectral camera in the spectral range of 400–1000 nm. Partial least squares regression (PLSR) models were developed to correlate the values of TC, TN, δ13C and δ15N, obtained from isotope ratio mass spectrometry method, with their spectral reflectance. The developed models provided acceptable predictions with high coefficient of determination (R2c) and low root mean square error (RMSEc) of calibration set for TC (R2c = 0.82; RMSEc = 1.08%), TN (R2c = 0.87; RMSEc = 0.02%), δ13C (R2c = 0.82; RMSEc = 0.27‰) and δ15N (R2c = 0.90; RMSEc = 0.29‰). The prediction abilities of the models were then evaluated using the spectra of an external test set (24 samples). The models provided excellent predictions with high R2t and ratio of performance to deviation (RPD) of test set for TC (R2t = 0.76; RPD = 2.02), TN (R2t = 0.86; RPD = 2.08), δ13C (R2t = 0.80; RPD = 2.00) and δ15N (R2t = 0.81; RPD = 1.94). The results indicated that the laboratory-based hyperspectral image analysis has the potential to predict soil TC, TN, δ13C and δ15N.

Assessment of Classifiers and Remote Sensing Features of Hyperspectral Imagery and Stereo-Photogrammetric Point Clouds for Recognition of Tree Species in a Forest Area of High Species Diversity

Recognition of tree species and geospatial information on tree species composition is essential for forest management. In this study, tree species recognition was examined using hyperspectral imagery from visible to near-infrared (VNIR) and short-wave infrared (SWIR) camera sensors in combination with a 3D photogrammetric canopy surface model based on RGB camera stereo-imagery. An arboretum with a diverse selection of 26 tree species from 14 genera was used as a test area. Aerial hyperspectral imagery and high spatial resolution photogrammetric color imagery were acquired from the test area using unmanned aerial vehicle (UAV) borne sensors. Hyperspectral imagery was processed to calibrated reflectance mosaics and was tested along with the mosaics based on original image digital number values (DN). Two alternative classifiers, a k nearest neighbor method (k-nn), combined with a genetic algorithm and a random forest method, were tested for predicting the tree species and genus, as well as for selecting an optimal set of remote sensing features for this task. The combination of VNIR, SWIR, and 3D features performed better than any of the data sets individually. Furthermore, the calibrated reflectance values performed better compared to uncorrected DN values. These trends were similar with both tested classifiers. Of the classifiers, the k-nn combined with the genetic algorithm provided consistently better results than the random forest algorithm. The best result was thus achieved using calibrated reflectance features from VNIR and SWIR imagery together with 3D point cloud features; the proportion of correctly-classified trees was 0.823 for tree species and 0.869 for tree genus.

Using laboratory-based hyperspectral imaging method to determine carbon functional group distributions in decomposing forest litterfall

Studying C functional group distributions in decomposing litterfall samples is one of the common methods of studying litterfall decomposition processes. However, the methods of studying the C functional group distributions, such as 13C NMR spectroscopy, are expensive and time consuming and new rapid and inexpensive technologies should be sought. Therefore, this study examined whether laboratory-based hyperspectral image analysis can be used to predict C functional group distributions in decomposing litterfall samples. Hyperspectral images were captured from ground decomposing litterfall samples in the visible to near infrared (VNIR) spectral range of 400–1000 nm. Partial least-square regression (PLSR) and artificial neural network (ANN) models were used to correlate the VNIR reflectance data measured from the litterfall samples with their C functional group distributions determined using 13C NMR spectroscopy. The results showed that alkyl-C, O,N-alkyl-C, di-O-alkyl-C1, di-O-alkyl-C2, aryl-C1, aryl-C2 and carboxyl derivatives could be acceptably predicted using the PLSR model, with R2 values of 0.72, 0.73, 0.71, 0.74, 0.76, 0.75 and 0.63 and ratio of prediction to deviation (RPD) values of 1.86, 1.82, 1.78, 1.71, 1.90, 1.76 and 1.43, respectively. Predicted O,N-alkyl-C, di-O-alkyl-C1, di-O-alkyl-C2, aryl-C1 and aryl-C2 using the ANN model provided R2 values of 0.62, 0.68, 0.69, 0.82 and 0.67 and the RPDs of 1.54, 1.76, 1.52, 2.10 and 1.72, respectively. With the exception of aryl-C1, the PLSR model was more reliable than the ANN model for predicting C functional group distributions given limited amount of training data. Neither the PLSR nor the ANN model could predict the carbohydrate-C and O-aryl-C acceptably. Overall, laboratory-based hyperspectral imaging in combination with the PLSR modelling can be recommended for the analysis of C functional group distribution in the decomposing forest litterfall samples.