Visible hyperspectral image (V-HSI) and thermal infrared hyperspectral image (TI-HSI) have been crucial data sources for land cover classification. V-HSI can directly provide the information of land surface, such as the shape, color, texture, and other features. TI-HSI contains rich long-wave spectral information, which can reflect the unique emission characteristics of ground objects in the thermal infrared spectral range. To fully leverage the advantages of V-HSI and TI-HSI while enhancing the classification accuracy, this paper proposes a self- and cross-attention enhanced transformer network (SCAET), integrated with convolutional neural network (CNN) for HSI classification. Initially, the proposed method employs a dual-branch spatial-spectral CNN (SS CNN) to extract spectral convolution features from V-HSI and TI-HSI respectively. Subsequently, a spectral feature mapping (SFM) module is proposed to perform feature transformation, extracting independent and interactive features of V-HSI and TI-HSI. Then, a self- and cross-attention interactive enhancement (SCIE) module is designed to extract deeper features and enhance the independent features by using the interactive features. Additionally, a self-projection mixing (SPM) module is formulated to promote feature interaction and improve the generalization capability of the model. To validate the effectiveness of the proposed network, extensive experiments are conducted on real-world datasets, and the results indicate that SCAET significantly outperforms current multi-source fusion networks.

The urban heat island (UHI) effect significantly impacts urban environments and quality of life, yet research comparing coastal and inland cities is relatively lacking. This study, using the MYD11A2 dataset, analyzes the spatiotemporal evolution of land surface temperature (LST) and the surface urban heat island intensity index (SUHIII) in Beijing (inland) and Dalian (coastal) from 2003 to 2023, exploring the driving factors from 2003 to 2018 and proposing mitigation strategies for similar cities. Key findings: (1) Beijing’s SUHIII is 0.45 °C higher than Dalian’s during summer days, while Dalian’s SUHIII is 0.24 °C stronger than Beijing’s during winter nights, likely due to Dalian’s maritime climate, which raises nighttime LSTs and intensifies the winter SUHIII. (2) Both cities show similar trends in LST and SUHIII, with fluctuations until 2010, an increase after 2011, and a peak in 2023, with the expansion of heat island areas occurring mainly in suburban regions. (3) From 2003 to 2018, TEMP is the primary factor promoting SUHIII, followed by ET and POP, with EVI as the main mitigating factor. Beijing’s PREP weakens SUHI, while Dalian’s PREP promotes it. Coastal cities should focus on water bodies and wetland planning to mitigate heat islands, especially in areas like Dalian which are affected by PREP.

The Huanjing-2A/2B (HJ-2A/2B) satellites are China’s next-generation environmental monitoring satellites, equipped with four visible light wide-swath charge-coupled device (CCD) sensors. These sensors enable the acquisition of 16-m multispectral imagery (16m-MSI) with a swath width of 800 km through field-of-view stitching. However, traditional vicarious calibration techniques are limited by their calibration frequency, making them insufficient for continuous monitoring requirements. To address this challenge, the present study proposes a spectral-angle difference correction-based cross-calibration approach, using the Landsat 8/9 Operational Land Imager (OLI) as the reference sensor to calibrate the HJ-2A/2B CCD sensors. This method improves both radiometric accuracy and temporal frequency. The study utilizes cloud-free image pairs of HJ-2A/2B CCD and Landsat 8/9 OLI, acquired simultaneously at the Dunhuang and Golmud calibration sites between 2021 and 2024, in combination with atmospheric parameters from the European Centre for Medium-Range Weather Forecasts (ECMWF) Reanalysis v5 (ERA5) dataset and historical ground-measured spectral reflectance data for cross-calibration. The methodology includes spatial matching and resampling of the image pairs, along with the identification of radiometrically stable homogeneous regions. To account for sensor viewing geometry differences, an observation-angle linear correction model is introduced. Spectral band adjustment factors (SBAFs) are also applied to correct for discrepancies in spectral response functions (SRFs) across sensors. Experimental results demonstrate that the cross-calibration coefficients differ by less than 10% compared to vicarious calibration results from the China Centre for Resources Satellite Data and Application (CRESDA). Additionally, using Sentinel-2 MSI as the reference sensor, the cross-calibration coefficients were independently validated through cross-validation. The results indicate that the radiometrically corrected HJ-2A/2B 16m-MSI CCD data, based on these coefficients, exhibit improved radiometric consistency with Sentinel-2 MSI observations. Further analysis shows that the cross-calibration method significantly enhances radiometric consistency across the HJ-2A/2B 16m-MSI CCD sensors, with radiometric response differences between CCD1 and CCD4 maintained below 3%. Error analysis quantifies the impact of atmospheric parameters and surface reflectance on calibration accuracy, with total uncertainty calculated. The proposed spectral-angle correction-based cross-calibration method not only improves calibration accuracy but also offers reliable technical support for long-term radiometric performance monitoring of the HJ-2A/2B 16m-MSI CCD sensors.

Pseudoinvariant calibration site (PICS) calibration is an important method for ensuring the on-orbit radiometric accuracy of Earth observation satellites. The proper selection of calibration sites is essential for ensuring the quality of quantitative remote sensing data and consistency across various sensors. This study presents a PICS selection method based on object analysis and a comprehensive evaluation of multidimensional characteristics, identifying qualified sites in the Golmud region of the Qaidam Basin. In this study, Landsat 8 Operational Land Imager (OLI) remote sensing data and related auxiliary data from 2014–2023 are used to comprehensively evaluate the calibration site from multiple dimensions, including spatiotemporal spectral consistency, site flatness, climatic characteristics, and trend stability. The results are as follows. 1) The proposed method selects sites with excellent attributes, with no significant correlation observed between spectral band consistency and spatiotemporal stability, indicating these should be treated as independent indicators. 2) Segmented objects based on the spatial neighborhood intensity constraint (SLIC) serve as units for site selection, which is suitable for PICS selection at various scales and effectively prevents spatial fragmentation issues linked to pixel-scale selection based on coefficient of variation (CV). Compared with the sliding window method, the selected sites are better in terms of spatiotemporal stability, with the mean spatiotemporal CV decreasing from 1.699$\%$ to 1.594$\%$. 3) The PICSs meeting the characteristic requirements are mainly distributed in the central Qaidam Basin and the northwestern area of Guolemude town. The CVs of both GRM$\_$1 and GRM$\_$2 are less than 1.6$\%$, with a dry climate, low cloud coverage and precipitation, and a uniform surface, making them suitable as supplementary sites for high-frequency calibration and correction.

Land surface temperature (LST) is a crucial variable in the Earth’s surface system, playing a key role in understanding the exchanges of material and energy between the surface and the atmosphere. Hyperspectral thermal infrared (TIR) data provide new opportunities for developing methods to retrieve LST from satellite observations. However, the typical physical hyperspectral TIR LST retrieval methods are limited by their reliance on accurate atmospheric correction and specific assumptions, which would introduce complexity and reduce applicability. To address these challenges, this study presents a novel LST retrieval framework that combines a Deep Residual Regression Network (DR2N) with the Iterative Spectrally Smooth Temperature and Emissivity Separation (ISSTES) algorithm, refined by Hampel filtering. In this framework, DR2N is trained on simulated data to efficiently retrieve atmospheric parameters, including upwelling radiance, downwelling radiance and transmissivity, and after that the refined ISSTES algorithm is applied to simulated data covering various surface types, yielding an overall RMSE of 1.89 K and a bias of -0.19 K. Subsequently, to further verify the performance of the proposed algorithm, LSTs over four study areas-Spain, North Africa, Hulunbuir, and the Yellow Sea are retrieved, and are compared with IASI L2 surface temperature products originating from the European Organization for the Exploitation of Meteorological Satellites (EUMETSAT), showing an RMSE of 0.51 K and a bias of -0.25 K. This confirms the efficacy of the proposed LST retrieval approach.