The retrieval of impervious surface information is a hot topic in remote sensing. However, researches on impervious surface retrieval from hyperspectral remote sensing imagery are rare. This paper illustrates a case study of information extraction from urban impervious surfaces based on hyperspectral remote sensing imagery that is intended to improve the image spectral resolution of impermeable materials. Fuzhou, Guangzhou, and Hangzhou were selected as test areas and EO-1 Hyperion images were used as data sources. The impervious surface features were retrieved from remote sensing images using linear spectral mixture analysis. A stepwise discriminant analysis was performed to select feature bands for impervious surface retrieval. A standard deviation analysis, correlation analysis, and principal component analysis were then carried out for each of those up to 158 valid Hyperion spectral bands. Eleven feature bands were selected using the stepwise discriminant analysis and a new image called Hyperion’ was formed. The impervious surface was then retrieved from Hyperion’. The results indicate that the extraction accuracy and coverage accuracy are high in all three test areas. Tests of eleven feature band combinations selected in different areas show very good representations of the band combinations in impervious surface retrieval, and can thus be used as optimal band combinations for impervious surface retrieval.

As the most important component of the urban surface, impervious surface has been recognized as a main indicator of urban heat island effect. However, the quantitative relationship between impervious surface area (ISA) and land surface temperature (LST) is still unclear as different studies achieved different results. Therefore, the present paper selected six cities from different geographical regions of China to further investigate the quantitative relationship between ISA and LST. The cities involved are Shanghai, Guangzhou, Beijing, Changsha, Lanzhou and Fuzhou. The ISAs of the six cities were derived from Landsat ETM+ images by using linear spectral mixture analysis and the LSTs of the cities were retrieved from the thermal infrared band of the images. Multivariate regression analysis with a large amount of samples was implemented to examine quantitative relationship between ISA and LST. Results show that the ISAs of the six cities are all positively correlated to the LST, but with an exponential relationship being the best fit model. This suggests that the area with high percent ISA could accelerate LST rise. The weaker transpiration evaporation and lower vegetation cover are the main factors contributing to the higher LST rise in the high percent ISA area.

Digital elevation models (DEMs) are foundational for critical tasks such as flood inundation simulation, disaster risk assessment, and ecosystem monitoring in coastal zones, yet their vertical accuracy is significantly compromised by complex terrain and surface characteristics. This study quantitatively decomposes the vertical errors of three 30 m global DEMs (COP30, NASADEM, and AW3D30) across the subtropical coastal region of Southeast China using ICESat-2 ATL08 data as a reference. By integrating an eXtreme Gradient Boosting (XGBoost) model with SHapley Additive exPlanations (SHAP), we successfully decoupled systematic biases from random noise. The results show that NASADEM achieved the lowest RMSE (7.775 m), followed by COP30 and AW3D30. While the Terrain Ruggedness Index (TRI) and categorically encoded Land Cover were identified as the universally dominant error drivers across all datasets, explainable analysis revealed distinct secondary mechanisms: X-band COP30 is notably susceptible to canopy height, exhibiting significant positive bias in forests exceeding 15 m; C-band NASADEM shows a systematic bias related to topographic position, typically overestimating ridges and underestimating valleys; and optical AW3D30 is significantly affected by stereo-matching errors. Furthermore, the analysis quantified a systematic error component of ~40%. These findings provide a data-driven basis for DEM selection and highlight that accuracy improvements should prioritize vegetation removal for radar DEMs and enhanced stereo-matching for optical models.

As an indispensable part of the marine ecosystem, the health status of the sea affects the stability and enhancement of the overall ecological function of the ocean. Clarifying the future land and sea utilization pattern and the impacts on the carbon stocks of island ecosystems is of great scientific value for maintaining marine ecological balance and promoting the sustainable development of the island ecosystem. Using Pingtan Island as an example, we simulate and predict changes in island utilization and carbon stocks for historical periods and multiple scenarios in 2030 via the PLUS-InVEST model and the marine biological carbon sink accounting method. The results show that (1) from 2006 to 2022, the carbon stock of Pingtan Island decreased by 7.218 × 10⁴ t, resulting in a cumulative economic loss of approximately USD 13.35 million; furthermore, from 2014 to 2018, the implementation of many reclamation and land reclamation projects led to a severe carbon stock loss of 6.634 × 10⁴ t. (2) By 2030, the projected carbon stock under the three different policy scenarios will be greater than that in 2022. The highest carbon stock of 595.373 × 10⁴ t will be found in the ecological protection scenario (EPS), which will be 4.270 × 10⁴ t more than that in 2022. With the strong carbon sequestration effect of the ocean, the total social carbon cost due to changes in island utilization is projected to decrease in 2030. (3) The factors driving changes in island utilization will vary in the design of different future scenarios. The results of this study not only provide a solid scientific basis for the sustainable development of island areas, but they also highlight the unique contribution of islands in the field of marine ecological conservation and carbon management, contributing to the realization of the dual-carbon goal.

Satellite-derived shorelines (SDS) reconstruct storm-scale beach change, but their accuracy degrades on macrotidal coasts. The prevailing storm-response and recovery framework, moreover, was built on microtidal, swell-dominated coasts. We combine CoastSat multi-mission shorelines (Sentinel-2, Landsat 7–9, 2015–2026) with EOT20 tidal correction, data-driven slope inversion and an expert-in-the-loop beach registry at five sites on the macrotidal (4–7 m) coast of Fujian, China, validate them against 19 sub-metre reference waterlines (18 yielding paired offsets), and apply them to 19 typhoons. Validation resolves the SDS error into a structure: unbiased near mean sea level, a slope-scaled landward bias of ~7–45 m toward low tide, and undefined below a beach-face/low-tide-flat break near −1.5 m, motivating a tide-gated protocol that cuts series scatter to 7–19 m. Typhoon retreats (median 15–16 m) vary systematically with landfall and embayment geometry, whereas offshore wave metrics and backshore hardening show no consistent event-scale control. Recovery is gated by the monsoon calendar: half-restored within weeks in summer, arrested through the northeast monsoon, and completed only the following March–June. This departs from the exponential recovery of mid-latitude coasts. The protocol and event catalogue provide a template for other macrotidal, typhoon-exposed coasts.

Fujian has one of China’s most tide-dominated and heavily engineered coasts, yet no segment-scale vulnerability assessment exists. We assessed 2449 km of mainland shoreline in 1 km segments using the classic six-variable coastal vulnerability index, from public data. Change rates came from thirty annual Landsat-based shorelines, giving a complete index for 2293 segments. The very high vulnerability class (defined by provincial quartiles) is predominantly concentrated in Putian (51.9 percent) and Quanzhou (42.8 percent). Those differences follow shoreline type. Eighty percent of the remaining 115 km of sandy shore is very high, against 53 percent of rocky shore in the low class. An exact decomposition gives each segment a dominant contributor from tide in Ningde to waves in Zhangzhou. Thirty percent of the 1.70 million people within 1 km of the coast live behind the very high quarter. Eleven sensitivity runs move individual segments, but none breaks the link to shoreline type, and Putian and Quanzhou lead even without the rate variable. The two published conventions for scoring tidal range rank different prefectures first, and each matches a different hazard. The storm-framed convention leads with Zhangzhou, where 99.7 percent of the province’s 2020 storm-surge damage fell.