Skip to Content
  • 4.3
    Impact Factor
  • 9.4
    CiteScore
  • 22 days
    Submission to First Decision
  • 3 days
    Acceptance to Publication

Remote Sensing

Remote Sensing is an international, peer-reviewed, open access journal about the science and application of remote sensing technology, published semimonthly online by MDPI. The Remote Sensing Society of Japan (RSSJ) and Japan Society of Photogrammetry and Remote Sensing (JSPRS) are affiliated with Remote Sensing and their members receive discounts on the article processing charge.

Get Alerted

Add your email address to receive forthcoming issues of this journal.

All Articles (44,204)

  • Article
  • Open Access

Large ancient landslides are widespread in tectonically active mountainous regions, and their reactivation poses a serious threat to communities in high-mountain gorge regions. However, reactivation mechanisms remain poorly understood for ancient landslides with locked segments and weak deformation responses to rainfall. This study investigated the deformation characteristics and reactivation mechanisms of the Bogexi ancient landslide in the Batang Fault Zone, eastern Tibetan Plateau, using high-resolution optical remote-sensing interpretation, field investigations, unmanned aerial vehicle photogrammetry, small baseline subset interferometric synthetic aperture radar (SBAS-InSAR) time-series deformation analysis, and numerical simulations. The results indicate that the landslide is approximately 2500 m long and 600–800 m wide, with an elevation difference of 1268 m and an estimated volume of 3.5–4.5 × 107 m3. It comprises two landslide bodies (I and II) separated by a central locked segment. SBAS-InSAR observations from 2017 to 2025 recorded the maximum cumulative displacement reached −2855 mm, and the maximum slope-parallel deformation rate was −362.5 mm/a. Deformation was concentrated in the central and frontal parts of landslide body I and throughout landslide body II. The central locked segment underwent creep at approximately 91.2 mm/a. These findings reveal a deformation pattern involving pushing from the rear section, resistance from the central locked segment, and traction at the toe. The landslide is dominated by long-term gravity-driven creep, and the locked segment controls the deformation. Numerical simulations indicated that seismic surface amplification may promote rupture propagation and coalescence within the locked segment. Under combined seismic and rainfall conditions, the maximum landslide displacement was predicted to reach 15 m, approximately 5.4 times that under natural conditions. The reactivation of the landslide can be divided into three stages: long-term creep, accelerated deformation driven by endogenic and exogenic geological processes, and instability following the failure of the locked segment. These findings would support identification and hazard assessment of such landslides in the alpine gorge regions of the Tibetan Plateau.

Remote Sens.

22 November 2026

Geological setting of the Bogexi landslide area and distribution of typical large landslides (optical remote-sensing imagery from Google Earth and World Imagery). (a) Tectonic setting of the study area. (b) Stratigraphic map of the landslide area; (c) remote sensing interpretation of the Bogexi ancient landslide; (d) remote sensing interpretation of the Zhama ancient landslide. The arrows in panels (c,d) indicate the direction of landslide movement.
  • Article
  • Open Access

Hyperspectral unmixing is a pivotal technique for analyzing mixed pixels in remote sensing imagery. However, the presence of spectral variability in crop hyperspectral images significantly compromises the accuracy of traditional unmixing methods. To address this challenge, this paper proposes a novel Endmember Adaptive Stacked Coding Network (EASCoN). The core of EASCoN is an innovative dual-branch architecture. The endmember encoding branch learns a pixel-dependent latent representation from the input spectrum and uses it to support adaptive endmember reconstruction, while the spatial–spectral branch extracts discriminative contextual features. A bidirectional cooperative optimization between the two branches is achieved through a spectral agent attention module and a weight-sharing mechanism, enabling mutual enhancement of spectrally adaptive features and spatial-contextual information. Furthermore, a comprehensive loss function incorporating multiple physical constraints is introduced to ensure the physical plausibility and spectral fidelity of the unmixing results. Extensive experiments conducted on the dedicated Wenchang and Wenjiao crop hyperspectral datasets, as well as the public Samson benchmark, demonstrate that EASCoN outperforms state-of-the-art methods across multiple evaluation metrics.

Remote Sens.

21 November 2026

Overall architecture of the proposed EASCoN model.
  • Article
  • Open Access

Chinese Mini SAR satellites, such as Dianjian-1, Shenqi-2, and the Huantian satellite (DP-15), are domestically developed high-resolution synthetic aperture radar (SAR) remote sensing satellites. With satellite masses of approximately 285–330 kg, these satellites are capable of rapid emergency observation and can acquire high-resolution surface information under cloudy and rainy conditions, providing critical data support for emergency flood monitoring and disaster assessment. Taking the Guangxi flood event on 6 July 2026, as a case study, this study utilizes high-resolution SAR imagery acquired by Shenqi-2 on 6 July 2026 (11 h after the dam failure) and Dianjian-1 on 7 July 2026 (25 h after the dam failure) to identify the flood extent and analyze its spatial characteristics. Sentinel-2 optical imagery and SAR imagery from the DP-15 (acquired on 7 July 2026, 24 h after the dam failure) are also used for cross-validation. The results show that the surfaces of the two reservoirs contracted rapidly on the day of the dam failure (6 July), while approximately 25.37 km2 of new inundated area emerged west of Jiajian–Futang Village in Yunbiao Town downstream. The floodwaters primarily propagated along river valleys, drainage channels, and relatively flat agricultural areas. The Sentinel-2 images obtained six days after the disaster still have about half of the spatial overlap. By 7 July, the combined area of the two reservoirs had decreased by a further 42.29%, while the downstream inundated area had receded to 11.46 km2. Building on these results, the flood-mapping capabilities of Dianjian-1, Shenqi-2, and Sentinel-1 were compared based on image-level land-cover details and backscatter profiles across representative land-cover types. The results indicate that Dianjian-1 and Shenqi-2 preserved water–land boundaries and small-scale water-body features more completely than Sentinel-1. The mean response ranges across representative water–land boundaries in the two high-resolution SAR datasets were approximately 17.5–20.5 m, whereas those of Sentinel-1 were approximately 3–4 times wider. For roads, the response ranges of Sentinel-1 were approximately 2–5 times wider than those of the high-resolution SAR datasets, indicating that high-resolution SAR satellites can capture spatial variations in land-surface backscatter within narrower spatial ranges. In densely built-up flood-affected areas, the X-band imagery exhibited more pronounced responses than the C-band imagery. Overall, small high-resolution SAR satellites demonstrate strong capabilities for rapid and fine-scale flood mapping, providing timely and detailed data support for emergency flood monitoring and disaster assessment.

Remote Sens.

21 November 2026

Study area overview. (a) Geographic location; (b) Aerial photograph of the dam breach at Liulan Reservoir; (c) Ground-level photograph of flood damage.
  • Article
  • Open Access

Pumice rafts from submarine volcanic eruptions can drift over long distances, disrupting navigation, fisheries, and coastal ecosystems. Their changing distribution calls for detection methods that can be applied repeatedly with limited manual intervention. We developed a method combining spectral discrimination and one-class Gaussian mixture models (GMMs) using observations of the 2021 Fukutoku-Oka-no-Ba eruption from the Medium Resolution Spectral Imager II (MERSI-II) aboard Fengyun-3D (FY-3D). The pumice index (PI) and four complementary indices (I1–I4) were calculated from four visible and near-infrared bands and jointly thresholded across each input image. Initial index thresholds were set manually, and pumice regions of interest in the 16 August and 12 September results were manually confirmed and used to fit a separate GMM for each date. The fitted models were then applied to the independent 22 September acquisition without updating their parameters or acceptance thresholds. Evaluation against visually interpreted reference labels covered 669,653 pixels within manually delineated assessment ROIs near known pumice rafts across the three dates. The spectral-index method achieved precision of 94.35–97.00%. On the 269,658 assessment pixels for 22 September, the GMM union achieved 97.02% precision, 74.45% recall, and a Cohen’s kappa of 0.84. Relative to spectral-index detection, precision and recall increased by 2.66 and 14.91 percentage points, respectively, with a relative reduction of 36.86% in the omission rate. Additional detections occurred along fragmented bands, patch margins, and scattered accumulations. The GMM-based workflow recovered additional reference pumice pixels missed by the spectral-index workflow while retaining high precision within the assessment ROIs. Following this manual initialization, the fixed models were applied to the later acquisition without image-specific retuning. These results support the use of FY-3D imagery for repeated pumice detection within the studied eruption and identify a further source of observations for monitoring volcanic hazards at sea.

Remote Sens.

21 November 2026

Location of Fukutoku-Oka-no-Ba volcano and the observed distribution of drifting pumice rafts. Adapted from Iskandar et al. [18].

Highly Accessed Articles

News & Conferences

Latest Issues

Open for Submission

Journal Sections

XFacebookLinkedIn
Remote Sens. - ISSN 2072-4292