Satoru Yamamoto

Hyperspectral (HS) sensors offer high spectral discrimination but generally limited spatial coverage, whereas multispectral (MS) sensors provide broad coverage with lower spectral detail. HS-MS complementary mapping aims to reduce coverage gaps in HS sensors by training classifiers within HS-MS overlap regions and applying them to surrounding MS-only areas. However, comprehensive comparisons across classifiers remain scarce in this complementary mapping setting. Furthermore, lightweight pixel-wise classifiers often produce spatially inconsistent predictions, whereas spatial–spectral deep learning models demand substantial computational resources. In this study, we propose a flexible HS-MS complementary mapping framework by systematically evaluating 13 classifiers across mineral and land-use/land-cover (LULC) mapping tasks and introducing class-adaptive uncertainty revocation (CAUR), a lightweight, classifier-independent post-processing module. Performance was evaluated via spatial holdout cross-validation within the HS-MS overlap area. When computational resources are sufficient, 3D convolutional neural networks (3D-CNN) achieve the highest accuracy. Conversely, lightweight models such as random forest (RF) and k-nearest neighbors (kNN) provide computationally efficient and robust baselines. Applying CAUR consistently improves spatial consistency and overall classification accuracy without model retraining, with the largest improvements observed for coarser-resolution HS reference data. These findings provide practical design guidelines for constructing efficient HS-MS complementary mapping pipelines tailored to application demands and sensor characteristics.

We performed spectral measurements on mixtures of ilmenite powder (μ m) and clinopyroxene or gabbro powder (μ m) to investigate the effect of ilmenite content on the spectral characteristics in the ultraviolet (UV), visible (Vis), and near-infrared (NIR) ranges. We found that increasing ilmenite content leads to (1) a darkening of Vis and NIR reflectance, (2) a decrease of the UV slope, (3) a weakening of band depths around 950–1050 nm and 1250–1300 nm, and (4) a neutralization of the Vis–NIR slope. For lunar mare regions, which have ∼7wt% ilmenite, it was estimated that the albedo in the Vis–NIR reflectance would decrease by 25%–29%, the Vis–NIR slope would decrease by 9%–26%, and the UV slope would decrease by 4%–5%. We also examined the effect of the addition of clinopyroxenes and gabbro on the spectral characteristics of ilmenite. The results indicate that most diagnostic features of ilmenite were masked, except for a distinct peak near 900 nm, which can thus serve as a reliable indicator of the presence of ilmenite in mixed compositions. Furthermore, we evaluated existing methods for estimating TiO ₂ content in lunar mare regions using spectral band ratios. The results indicate that ratios that incorporate UV reflectance are more sensitive to ilmenite content than those that incorporate reflectance only at Vis and/or NIR wavelengths longer than 415 nm. This suggests that UV reflectance is more reliable for the estimation of ilmenite abundance on the lunar surface.

We studied the global distribution and geological features of lunar surface sites whose spectra indicate an ilmenite‐rich composition. Hyperspectral data obtained by the Kaguya Spectral Profiler were used for data mining to identify diagnostic features of a 1‐ and 2‐m spectral reflectance of ilmenite, revealing the global distribution of sites showing ilmenite‐rich spectra. The results show that regions with ilmenite‐rich spectra are concentrated at the margins of impact basins on the lunar nearside, whereas no such regions are identified in the Feldspathic Highland Terrain or the South Pole‐Aitken basin. Using multiband images and a digital terrain model obtained by the Kaguya Multiband Imager and Terrain Camera, we examined the geological features of each site showing ilmenite‐rich spectra and found that most of the sites are distributed on pyroclastic deposits overlying highland materials. Spectra interpreted as glass‐rich material are prevalent in and around areas having ilmenite‐rich spectra. However, sites showing ilmenite‐rich spectra do not correspond to mare regions with ‐rich basalts. These results may indicate that the concentration of ilmenite in pyroclastic deposits is high enough to exhibit diagnostic features of 1‐ and 2‐m spectral reflectance of ilmenite, whereas the concentration in mare regions with ‐rich basalt is not. Since pyroclastic deposits are expected to be extensive, deep unconsolidated deposits of relatively block‐free debris, resulting in high processing efficiency in the hydrogen reduction processes, our data may be useful for developing an efficient exploration strategy for ilmenite as a lunar resource.

The hyperspectral imager suite (HISUI) is a recently commissioned spaceborne hyperspectral sensor that is expected to provide higher-quality information for terrestrial and aquatic monitoring at a global scale. To maximize the utility of the detailed spectra observed by HISUI, calibration and validation activities have been intensively conducted for the past several years. However, research on long-term monitoring of sensor radiometric characteristics and checking inter-band radiometric consistency is still limited. This study provided the first results of cross- and inter-band calibrations using five scenes acquired over a desert calibration site during 2021–2023, and calculated relative cross-calibration coefficients (RCCCs) to calibrate all the HISUI bands. To make full use of the limited amount of data archived over the calibration site, off-nadir scenes obtained by wide-swath counterpart sensors [i.e., the moderate resolution imaging spectroradiometer (MODIS)] were used for the cross-calibration, with correction by a bidirectional reflectance distribution function. The obtained RCCCs successfully mitigated the inter-band radiometric variability of HISUI data, which was consistent with the in situ surface reflectance data observed by the radiometric calibration network (RadCalNet). Time-series analysis further revealed a notable discrepancy between the visible to near-infrared and short-wave infrared (SWIR) spectral domains after the International Space Station (ISS) maneuvers in spring 2022. This article is also a prompt report of this phenomenon and will soon contribute to HISUI radiometric calibration, particularly after the maneuver events.

This paper reports on the long-term variation of the spectral smile and the full-width at half-maximum (FWHM) of the sensor response function for the Hyperspectral Imager Suite (HISUI) onboard the International Space Station (ISS). HISUI is an optical hyperspectral imager comprising two subsystems: a visible-and-near-infrared (VNIR) system covering 400 to 980 nm and a short-wave infrared (SWIR) system covering 895 to 2481 nm. On the basis of atmospheric corrections for actual observation images, we found that the characteristics of the spectral smile and the FWHM can be classified into five major periods. The results show that the first two transitions of the periods are due to updates of the smile correction table used in HISUI and that the last one is related to the occurrence of vertical stripes in the HISUI images. However, two unexpected changes are observed. One is a sudden increase in the spectral smile in the SWIR data to 2.0–2.1 nm on 30 September 2021, which was just 3 days after the second update of the smile correction table. The other is a sudden improvement on 5 January 2022, showing almost zero spectral smile and a nearly constant FWHM. Because neither sudden change is associated with HISUI or ISS operations/events, the optical and electrical systems of HISUI might exhibit instability, leading to sporadic changes of the spectral smile and FWHM. The findings of this study therefore suggest that, although the post-launch spectral smile correction has been completed, continued long-term evaluation of the spectral stability is important.

We studied the lunar mantle composition based on the characteristics of low‐Ca pyroxene (LCP)‐rich and olivine‐rich rocks exposed on the lunar surface. Using data mining with SELENE (Kaguya) hyperspectral data, we found 531 sites with spectra dominated by an ultramafic LCP end‐member mineral, most of which are located in the Imbrium basin and the South Pole‐Aitken (SPA) basin. Stratigraphic analysis of each site revealed that LCP‐rich rock bodies are exposed on fresh geological features that are less affected by space‐weathering, such as steep slopes at peaks, and crater walls and ejecta deposits at smaller craters. We also found that, in the SPA and Imbrium basins, LCP‐rich bodies are more numerous and more widely distributed than olivine‐rich rocks, suggesting that LCP‐rich materials deep in the mantle were excavated during the formation of these huge basins. However, olivine‐rich rocks were abundant, and no LCP‐rich rocks were found in the Moscoviense, Crisium, and Humboldtianum basins, which are known to have almost‐zero crustal thicknesses, indicative of mantle excavation. Thus, our results suggest that the composition of rocks derived from the lunar mantle varies with the impact basin. Such a difference might indicate a layered structure of mantle composition, with the olivine‐rich upper mantle overlying the LCP‐rich mantle, a horizontal heterogeneity in the mantle composition, a regional heterogeneity in early lunar basaltic magmatism, or an impact melt origin. The layered structure or horizontal heterogeneity might have resulted from a mantle overturn caused by gravitational instability in the early stages of the lunar magma ocean.

This paper reports on the initial analysis of spectral smile calibration of the Hyperspectral Imager Suite (HISUI) onboard the International Space Station, which has been continuously acquiring data since September 4, 2020. HISUI is an optical hyperspectral imager consisting of two subsystems: VNIR covering 400 to 980nm at intervals of 10 nm, and SWIR covering 895 to 2481nm at 12.5nm intervals. Based on the atmospheric correction for actual observation images, we assessed cross-track dependences of the wavelength deviation (spectral smile) and the full-width at half-maximum (FWHM) of the HISUI response function. We found that significant spectral smile was observed, with maximum variations of 1.8nm in VNIR and 4.3–4.5nm in SWIR. In addition, the cross-track variation of FWHM was observed with maximum variations of 5.0nm for VNIR and 2.5–3.5nm for SWIR.We used the results to model the smile functions to update a smile correction table in the internal calibration system of HISUI. Then, we evaluated how the smile functions reduce the spectral smile in the data acquired after the update on September 27, 2021.We confirmed that VNIR showed a nearly flat profile within 0.25nm with a nearly constant FWHM. For SWIR, although a slight amount of spectral smile and a variation of FWHM were still observed partly due to wavelength dependence in the spectral smile, the spectral smile was reduced to <~2.2 nm. This study demonstrated that wavelength calibration using actual observation images for ground surfaces is important for the characterization of hyperspectral sensors.

We examined the global distribution and geological context of lunar sites where olivine‐ and plagioclase‐rich materials co‐exist. These sites are areas showing plagioclase‐rich spectra adjacent to areas showing olivine‐rich spectra, and they extend over several hundreds of meters on the lunar surface. From an analysis of the high‐spatial‐resolution data obtained from the SELENE (Kaguya) multiband imager, we identified 14 co‐existing occurrences among 49 olivine‐rich sites located around large impact basins. We found that the co‐existing occurrences are limited to a few sites among the olivine‐rich sites in the Moscoviense, Crisium, and Imbrium basins, while six of seven olivine‐rich sites in the Schrödinger basin show co‐existing occurrence. The geological features between the co‐existing occurrences and the olivine‐rich sites without plagioclase‐rich materials showed no clear difference, but both types of site are found at fresh geological features located around the peak rings of impact basins. These results may suggest that the co‐existing occurrences are closely related to the material heterogeneity of the peak ring regions. The heterogeneity could be formed by mixing of the olivine‐rich mantle and the anorthosite‐rich crust during impact basin formation, or it might be a reflection of the compositional heterogeneity of the lunar lower crust. For future sample return missions, the detailed information on the co‐existing occurrences presented here might be useful for a sampling strategy to select a “one‐stop” site that provides important information for the overall characterization of lunar mantle and crustal materials.