An experimental comparison between KELM and CART for crop classification using Landsat-8 OLI data

The operational land imager (OLI) is the latest instrument in the Landsat series of satellite imagery, which officially began normal operations on 30 May 2013. The OLI includes two bands that are not on the thematic mapper series of sensors aboard Landsat-5 and 7; a cirrus band and a coastal/aerosol band. This paper compares the classification and regression tree and the kernel-based extreme learning machine (KELM) for mapping crops in Hokkaido, Japan, using OLI data, except the cirrus band and the pan band. The OLI data acquired on 8 July 2013 was used for crop classification of beans, beets, grassland, maize, potatoes and winter wheat. The KELM algorithm performed better in this study and achieved overall accuracies of 90.1%. According to the Jeffries–Matusita (J–M) distances, the short wavelength infrared band provides the greater contribution (the highest value was observed for band 6 in OLI data).     

Metamorphic Environments around the Cretaceous Kurihashi Pluton in the Kitakami Mountains, NE Japan: in Comparison to the Ganidake Pluton - Kamaishi Skarn Metasomatism

Abstract. Many granitic plutons of Early Cretaceous age are intruded on various scales in the Kitakami Mountains. The stock‐type Ganidake pluton accompanies enormous Fe‐Cu mineralization of the Kamaishi deposits, whereas the Kurihashi pluton accompanies less mineralization. To elucidate the cause of these differences, the metamorphic conditions and redox state of the contact metamorphic aureole around the Kurihashi pluton have been examined by the petrochemical study and gas analysis of the metamorphic rocks. A typical mineral assemblage in the pelitic rocks in the lowest‐grade part is biotite‐muscovite‐chlorite‐quartz‐plagio‐clase‐graphite, which occur more than 2 km away from the contact point with the Kurihashi pluton. Graphite disappears at the 1550 m point, and cordierite and garnet appear in the middle and highest‐grade parts, respectively. A typical mineral assemblage in the tuffaceous rocks in the lower‐grade part is chlorite‐actinolite‐biotite‐quartz‐plagioclase. Actinolite changes into hornblende near to the pluton. The CO₂/CH₄ ratios obtained in measurements by gas chromatography exceeds 100 in the pelitic rocks at the contact point with the pluton. The ratios decrease and become less than 0.1 with distance from the pluton. Equilibrium temperatures calculated from a garnet‐biotite pair in the pelitic rock and a hornblende‐plagioclase pair in the tuffaceous rock are 640d? and 681 d?C at the contact point, respectively. The log?o₂ values among these metamorphic aureoles estimated from the CO₂/CH₄ ratios are slightly lower than the FMQ‐buffer. Redox states of the contact metamorphic aureole are kept in an intermediate condition between oxidized magma of the Kurihashi pluton and graphite‐bearing pelitic country rocks. Judging from these metamorphic conditions around the Kurihashi pluton and from the re‐evaluation of the previous knowledge about contact metasomatism around the Ganidake pluton, the Kurihashi metamorphism has occurred at higher temperatures and dry conditions than the Ganidake metasomatism. These differences in the metamorphic conditions and presence or absence of a large limestone mass around the pluton might be the principal reasons why the Kurihashi pluton accompanies less mineralization and the Ganidake pluton accompanies gigantic Kamaishi skarn mineralization.     

Hydration of rhyolitic glass during weathering as characterized by IR microspectroscopy

The mechanism and rate of hydration of rhyolitic glass during weathering were studied. Doubly polished thin sections of two rhyolites with different duration of weathering (Ohsawa lava: 26,000 yr, Awanomikoto lava: 52,000 yr) were prepared. Optical microscope observation showed that altered layers had developed along the glass surfaces. IR spectral line profile analysis was conducted on the glass sections from the surface to the interior for a length of 250 μm and the contents of molecular H₂O (H₂O_m), OH species (OH) and total water (H₂O_t) were determined. The diffusion profile of H₂O_m in Ohsawa lava extends beyond the layer observed by optical microscope. The content of H₂O_m in the hydrated region is much higher than that of OH species. Thus, the reaction from H₂O_m to OH appears to be little and H₂O_m is the dominant water species moving into the glass during weathering. Based on the concentration profiles, the diffusion coefficients of H₂O_m(D_H2Om) and H₂O_t(D_H2Ot) were determined to be 2.8 × 10⁻¹⁰ and 3.4 × 10⁻¹⁰ μm² s⁻¹ for Ohsawa lava, and 5.2 × 10⁻¹¹ and 4.1 × 10⁻¹¹ μm² s⁻¹ for Awanomikoto lava, respectively. The obtained D_H2Om during weathering are more than 2-3 orders of magnitude larger than the diffusion coefficient at ∼20 °C that is extrapolated from the diffusivity data for >400 °C. This might suggest that the mechanism of water transport is different at weathering conditions and >400 °C.     

Oxygen isotope thermometry using quartz inclusions in garnet

Oxygen isotope ratios of quartz inclusions (QI) within garnet from granulite and amphibolite facies gneisses in the Adirondack Mountains, NY were analysed and used to determine metamorphic temperatures. Primary QI for eight of 12 samples have δ¹⁸O values significantly lower than matrix quartz (MQ). The primary QI retain δ¹⁸O values representative of thermal conditions during garnet crystallization, whereas the δ¹⁸O values of MQ were raised by diffusive exchange with other matrix minerals (e.g. mica and feldspar) during cooling. The δ¹⁸O differences between QI and MQ show that garnet (a mineral with slow diffusion of oxygen) can armour QI from isotopic exchange with surrounding matrix, even during slow cooling. These differences between δ¹⁸O in MQ and QI can further be used to test cooling rates by Fast Grain Boundary diffusion modelling. Criteria for identifying QI that preserve primary compositions and are suitable for thermometry were developed based on comparative tests. Relations between δ¹⁸O and inclusion size, distance of inclusion to host–garnet rim, core–rim zonation of individual inclusions, and presence or absence of petrological features (healed cracks in QI, inclusions in contact with garnet cracks lined by secondary minerals, and secondary minerals along the inclusion grain boundary) were investigated. In this study, 61% of QI preserve primary δ¹⁸O and 39% were associated with features that were linked to reset δ¹⁸O values. If δ¹⁸O in garnet is homogeneous and inclusions are removed, laser‐fluorination δ¹⁸O values of bulk garnet are more precise, more accurate, and best for thermometry. Intragrain δ¹⁸O(Grt) profiles measured in situ by ion microprobe show no δ¹⁸O zonation. Almandine–rich garnet (Alm_60–75) from each sample was measured by laser‐fluorination mass‐spectrometry (LF‐MS) for δ¹⁸O and compared with ion microprobe measurements of δ¹⁸O in QI for thermometry. The Δ¹⁸O(Qz–Grt) values for Adirondack samples range from 2.66 to 3.24‰, corresponding to temperatures of 640–740 °C (A[Qz–Alm] = 2.71). Out of 12 samples that were used for thermometry, nine are consistent with previous estimates of peak temperature (625–800 °C) based on petrological and carbon–isotope thermometry for regional granulite and upper amphibolite facies metamorphism. The three samples that disagree with independent thermometry for peak metamorphism are from the anorthosite–mangerite–charnockite–granite suite in the central Adirondacks and yield temperatures of 640–665 °C, ~100 °C lower than previous estimates. These low temperatures could be interpreted as thermal conditions during late (post‐peak) crystallization of garnet on the retrograde path.     

The Tiger Sulfide Chimney,Yonaguni Knoll IV Hydrothermal Field,Southern Okinawa Trough,Japan: The First Reported Occurrence of Pt–Cu–Fe‐Bearing Bismuthinite and Sn‐Bearing Chalcopyrite in an Active Seafloor Hydrothermal System

A sulfide chimney ore sampled from the flank of the active Tiger vent area in the Yonaguni Knoll IV hydrothermal field, south Okinawa trough, consists of anhydrite, pyrite, sphalerite, galena, chalcopyrite and bismuthinite. Electron microprobe analysis indicates that the chalcopyrite contains up to 2.4 wt% Sn, whereas bismuthinite contains up to 1.7 wt% Pt, 0.8 wt% Cu and 0.5 wt% Fe. The Sn‐rich chalcopyrite and Pt–Cu–Fe‐bearing bismuthinite are the first reported occurrence of such minerals in an active submarine hydrothermal system. The results confirm that Sn enters the chalcopyrite as a solid solution towards stannite by the coupled substitution of Sn⁴⁺Fe²⁺ for Fe³⁺Fe³⁺, whereas Pt, Cu and Fe enter the bismuthinite structure as a solid solution during rapid nucleation. The fluid inclusions homogenization temperatures in anhydrite (220–310°C) and measured end‐member temperature of the vent fluids on‐site (325°C) indicate that Sn‐bearing chalcopyrite and Pt–Cu–Fe‐bearing bismuthinite express the original composition of the minerals that precipitated as metastable phases at a temperature above 300°C. The result observed in this study implies that sulfides in ancient volcanogenic massive sulfide deposits have similar trace element distribution during nucleation but it is remobilised during diagenesis, metamorphism or supergene enrichment processes.     

Dehydration kinetics of antigorite using in situ high-temperature infrared microspectroscopy

The dehydration kinetics of serpentine was investigated using in situ high-temperature infrared microspectroscopy. The analyzed antigorite samples at room temperature show relatively sharp bands at around 3,655–3,660 cm^?1 (band 1), 3,570–3,595 cm^?1 (band 2), and 3,450–3,510 cm^?1 (band 3). Band 1 corresponds to the Mg–OH bond, and bands 2 and 3 correspond to OH associated with the substitution of Al for Si. Isothermal kinetic heating experiments at temperatures ranging from 625 to 700 °C showed a systematic decrease of the OH band absorbance with heating duration. The one-dimensional diffusion was found to provide the best fit to the experimental data, and diffusion coefficients were determined with activation energies of 219 ± 37 kJ mol^?1 for the total water band area, 245 ± 46 kJ mol^?1 for band 1, 243 ± 57 kJ mol^?1 for band 2, and 256 ± 53 kJ mol^?1 for band 3. The results indicate that the dehydration process is controlled by one-dimensional diffusion through the tetrahedral geometry of serpentine. Fluid production rates during antigorite dehydration were calculated from kinetic data and range from 3 × 10^?4 to 3 × 10^?5  $ {\text{m}}_{\text{fluid}}^{ 3} \,{\text{m}}_{\text{rock}}^{ - 3} \,{\text{s}}^{ - 1} $ . The rates are high enough to provoke hydraulic rupture, since the relaxation rates of rocks are much lower than these values. The results suggest that the rapid dehydration of antigorite can trigger an intermediate-depth earthquake associated with a subducting slab.     

Dehydration kinetics of muscovite by in situ infrared microspectroscopy

Dehydration behavior of muscovite flake was investigated at 760–860°C by using in situ high-temperature IR microspectroscopy for the OH absorption band around 3,620 cm⁻¹. Isothermal kinetic heating experiments at each temperature gave detailed decrease curves of the OH band area with time. These curves have been simulated by the first and second order reactions or mono- and two-dimensional diffusion processes. The mono-dimensional diffusion was found to give the best fit to the experimental data and apparent diffusion coefficients D were determined at 760–860°C with the activation energy of 290 ± 20 kJ/mol. The apparent diffusion coefficients D varied with the sample thickness L. This variation can be explained by an m layers model with a unit length of L′ with a constant diffusion coefficient D′. Therefore, the dehydration process might be rate-limited by mono-dimensional diffusion through tetrahedral silicate sheet perpendicular to (001) planes of muscovite with a unit length of L′.     

Assessing the suitability of data from Sentinel-1A and 2A for crop classification

Sentinel-1A C-SAR and Sentinel-2A MultiSpectral Instrument (MSI) provide data applicable to the remote identification of crop type. In this study, six crop types (beans, beetroot, grass, maize, potato, and winter wheat) were identified using five C-SAR images and one MSI image acquired during the 2016 growing season. To assess the potential for accurate crop classification with existing supervised learning models, the four different approaches namely kernel-based extreme learning machine (KELM), multilayer feedforward neural networks, random forests, and support vector machine were compared. Algorithm hyperparameters were tuned using Bayesian optimization. Overall, KELM yielded the highest performance, achieving an overall classification accuracy of 96.8%. Evaluation of the sensitivity of classification models and relative importance of data types using data-based sensitivity analysis showed that the set of VV polarization data acquired on 24 July (Sentinel-1A) and band 4 data (Sentinel-2A) had the greatest potential for use in crop classification.     

Oxygen and carbon isotope records of cultured freshwater pearl mussel <Emphasis Type="Italic">Hyriopsis</Emphasis> sp. shell from Lake Kasumigaura,Japan

We present oxygen and carbon isotope ratios and the morphological structure of the cultured freshwater pearl mussel (Hyriopsis sp., Unionidae) shell and pearl. The number of first-order fluctuations of δ¹⁸O of the outer shell layer along the maximum growth axis was consistent with the number of cultured years. The dominant factor controlling annual δ¹⁸O fluctuations was water temperature with a minor contribution from the variation in δ¹⁸O of ambient water, especially during the rainy season. The δ¹³C values were approximately constant throughout the life of the mussel, suggesting that the contributions of body size to δ¹³C of the shell were minor. We observed nine distinct disturbance rings on the outer surface of the shell. Five rings coincided with the five winter peaks of the δ¹⁸O profile, indicating winter growth cessation below approximately 10°C, probably because of either inactive growth at low water temperatures or reproduction. Summer disturbance rings were not observed in all years. Moreover, some summer rings showed discontinuity in the inner structure. These findings suggest that summer growth cessation may be caused by occasional events such as heavy rains, as the decrease of dissolved oxygen concentration. The δ¹⁸O profile and shell structures indicated that shell aragonite was precipitated at close to equilibrium conditions with respect to the oxygen isotope composition of the ambient water. Hyriopsis sp. shells can potentially be used for reconstruction of past hydrologic conditions. The δ¹⁸O of a pearl indicated that calcification occurred over a temperature range of at least 13–23°C. The optimal temperature for pearl calcification in this species is lower than that for marine pearl calcification.     

Search for extinct Cl: Vigarano CAIs, the Pink Angel from Allende, and a Ningqiang chondrule

We have searched for excesses of ³⁶S derived from the decay of extinct ³⁶Cl in sodalite, a secondary Cl-rich mineral, in Ca-Al-rich inclusions (CAIs) from the Vigarano and Allende CV3 chondrites and in a chondrule from the Ningqiang carbonaceous chondrite. The presence of sodalite in two CAIs from Vigarano and its absence from surrounding CAI fragments suggests sodalite formation after CAI fragmentation. As for sodalite in the Allende Pink Angel CAI, oxygen isotopic compositions have been interpreted as indicative of high temperature interactions, thus suggesting formation prior to accretion to the parent body, probably in a nebular setting. Sodalite in the Ningqiang chondrule is considered to have formed via alkali-Ca exchange, which is believed to have occurred before accretion to the parent body.Sodalites in the Vigarano CAIs and in the Ningqiang chondrule show no clear evidence for the presence of radiogenic ³⁶S. The inferred 2σ upper limits for ³⁶Cl/³⁵Cl at the time of sodalite formation are 1.6 × 10⁻⁶ (Vigarano CAIs) and 3.3 × 10⁻⁶ (Ningqiang chondrule), respectively. In the Pink Angel CAI sodalite exhibits small ³⁶S excesses which weakly correlate with ³⁵Cl/³⁴S ratios. The inferred ³⁶Cl/³⁵Cl ratio of (1.8 ± 2.2) × 10⁻⁶ (2σ error) is lower than that found by Hsu et al. [Hsu, W., Guan, Y., Leshin, L. A., Ushikubo, T. and Wasserburg, G. J. (2006) A late episode of irradiation in the early solar system: Evidence from extinct ³⁶Cl and ²⁶Al in meteorites. Astrophys. J. 640, 525-529], thus indicative of heterogeneous distribution of ³⁶Cl in this CAI. Spallation reactions induced by energetic particles from the young Sun are suggested for the origin of ³⁶Cl, similar to the case of ¹⁰Be. While ¹⁰Be appears to be present in roughly equal abundance in all studied CAIs, our study indicates the level of ³⁶Cl abundances to be variable so that there seems to be no simple relationship between ¹⁰Be and ³⁶Cl. This would be expected if trapped cosmic rays rather than Early Solar System spallation were the dominant source of ¹⁰Be in the Early Solar System, since their contribution to ³⁶Cl would have been tiny.If the variability of ³⁶Cl abundances is caused by temporal differences in the alteration that formed sodalite, sodalite in the Vigarano CAIs and in the Ningqiang chondrule may have formed ?0.5 and ?0.2 Ma after formation of the sodalite in the Ningqiang CAI analyzed by Lin et al. [Lin, Y., Guan, Y., Leshin, L. A., Ouyang, Z. and Wang, D. (2005) Short-lived chlorine-36 in a Ca- and Al-rich inclusion from the Ningqiang carbonaceous chondrite. Proc. Natl. Acad. Sci. USA 102, 1306-1311]. The Pink Angel sodalite that we analyzed may have formed ∼0.3 Ma after formation of the sodalite in the Pink Angel analyzed by Hsu et al. [Hsu, W., Guan, Y., Leshin, L. A., Ushikubo, T. and Wasserburg G. J. (2006) A late episode of irradiation in the early solar system: Evidence from extinct ³⁶Cl and ²⁶Al in meteorites. Astrophys. J. 640, 525-529]. The estimated temporal differences suggest that alteration extended over at least 0.5 Ma. If previous works showing very low initial ³⁶Cl/³⁵Cl in the Allende CAIs and a H3 chondrite CAI are also considered, then alteration processes may have been more than 1.7 Ma.