The rapid advancement of remote sensing technology has given rise to numerous global- and regional-scale medium- to high-resolution land cover (LC) datasets, making significant contributions to the exploration of worldwide environmental shifts and the sustainable governance of natural resources. Nonetheless, owing to the inherent uncertainties embedded within remote sensing imagery, LC datasets inevitably exhibit inaccuracies. In this study, a local accuracy assessment of LC datasets in Southwest China was conducted. The datasets utilized in our analysis include ESA WorldCover, CLCD, Esri Land Cover, CRLC, FROM-GLC10, GLC_FCS30, GlobeLand30, and SinoLC-1. This study employed a sampling approach that combines proportional allocation and stratified random sampling (SRS) to gather sample points and compute confusion matrices to validate eight LC products. The local accuracy of the eight LC maps differs significantly from the overall accuracy provided by the original authors in Southwest China. ESA WorldCover and CLCD demonstrate higher local accuracy than other products in Southwest China, with their overall accuracy (OA) values being 87.1% and 85.48%, respectively. Simultaneously, we computed the area for each LC map based on categories, quantifying uncertainty through the reporting of confidence intervals for both accuracy and area parameters. This study aims to validate and compare eight LC datasets and assess precision and area of diverse spatial resolution datasets for mapping and monitoring across Southwest China.

Eucalyptus plantations are one of the primary artificial forests in southern China, experiencing rapid expansion in recent years due to their significant socio-economic benefits. This expansion has raised concerns about the ecological environment, necessitating accurate mapping of eucalyptus plantations. In this study, the phenological characteristics of eucalyptus plantations were utilized as the primary classification basis. Long-term time series Landsat and Sentinel-2 data from 2000 to 2022 were rigorously preprocessed pixel by pixel using the Google Earth Engine (GEE) platform to obtain high-quality observation data. The Continuous Change Detection and Classification (CCDC) algorithm was employed to fit the multi-year observation data with harmonic curves, utilizing parameters such as normalized intercept, slope, phase, and amplitude of the fitted curves to characterize the phenological features of vegetation. A total of 127 phenological indices were generated using the Normalized Burn Ratio (NBR), Normalized Difference Fractional Index (NDFI), and six spectral bands, with the top 20 contributing indices selected as input variables for the random forest algorithm to obtain preliminary classification results. Subsequently, eucalyptus plantation rotation features and the Simple Non-Iterative Clustering (SNIC) superpixel segmentation algorithm were employed to filter the results, enhancing the accuracy of the identification results. The producer’s accuracy, user’s accuracy, and overall accuracy of the eucalyptus plantation map for the year 2020 were found to be 96.67%, 89.23%, and 95.83%, respectively, with a total area accuracy of 94.39%. Accurate mapping of eucalyptus plantations provides essential information and evidence for ecological environment protection and the formulation of carbon-neutral strategies.

Land cover products are essential tools in environmental and ecological research. However, limited attention has been paid to their data quality issues. Many existing products suffer from pronounced spatiotemporal inconsistencies, characterized by frequent and repetitive classification fluctuations in specific regions and years, which substantially compromise the accuracy of analyses and models that rely on them. To address these challenges, this study introduces a method for deriving spatiotemporally stable samples to support high-precision land cover classification. The approach integrates national and regional land-use policies to assess temporal stability and incorporates advanced time-series processing techniques together with innovative vegetation indices to facilitate effective sample reuse. Experimental results show that this method markedly improves classification accuracy across vegetation types and reduces the extent of areas prone to frequent land-cover changes by 22.64%. Compared with existing products of similar spatial resolution, our approach achieves an overall classification accuracy of 91.1%, providing stable, high-quality input data that underpin precise and reliable regional-scale environmental and ecological modeling.