Nighttime light (NTL) remote sensing has emerged as a powerful tool in various fields such as urban expansion, socio-economic estimation, light pollution, and energy domains. However, current annual NTL products suffer from several critical limitations, including poor consistency, severe background noise, and limited comparability. These issues have significantly interfered with the research of long-term NTL trends and diminished the accuracy of related findings. Therefore, this study developed a de-difference smoothing algorithm for producing high-quality annual NTL products based on monthly National Polar-orbiting Partnership Visible Infrared Imaging Radiometer Suite (NPP-VIIRS) NTL data. It enabled the construction of a continuous global high-quality NTL dataset, named the De-Difference Smoothed Nighttime Light (DDSNL), covering the period from 2012 to 2023. Comparative analyses were conducted to validate the accuracy and availability of the DDSNL product against the benchmark EOG NPP-VIIRS and NPP-VIIRS-like NTL datasets. The results showed that DDSNL products had strong correlation with the NTL distribution of EOG NPP-VIIRS, but little correlation with NPP-VIIRS-like. Notably, DDSNL demonstrated better background noise reduction and higher separability between NTL and non-NTL areas compared to EOG NPP-VIIRS NTL. In contrast to the complete exclusion of background in NPP-VIIRS-Like, the retention of background values in DDSNL leads to more reasonable representation in the urban fringes. In the analysis of NTL changes matching impervious surface changes, the DDSNL product demonstrated the least interference from noise, resulting in the smallest segmentation threshold and the highest matching accuracy. This indirectly demonstrates the spatial and temporal consistency of the annual DDSNL product, ensuring its reliability in change detection-related studies. The annual DDSNL product developed in this research exhibits high fidelity, strong consistency, and improved comparability, and can provide reliable data reference for applications in electrification and urban studies.
Forests and grasslands are crucial components of terrestrial ecosystems. Their stability and health play an essential role in maintaining ecological balance. However, with the intensification of human activities, these ecosystems are facing increasing risks of degradation. Forest and grassland change detection is vital for monitoring and management by governmental agencies, requiring reliable and automated detection models. Currently, most deep learning-based change detection models face significant limitations in achieving precise detection of forest and grassland changes. These challenges are particularly evident in addressing incomplete segmentation objects and accurately discriminating hard/challenging samples, highlighting substantial room for further research and improvement. To address these challenges, we have developed MCACD. This model effectively preserves detailed information by combining high-resolution feature maps with dynamic upsampling segmentation heads. It also integrates an efficient feature fusion module, designed by us, to enable the integration of dual-temporal features. Additionally, we introduce a highly efficient multi-scale cascade fully convolutional attention decoder, which progressively enhances spatial features at multiple scales and resolutions. A deep supervision strategy within the decoder further optimizes the feature learning process. Experimental results show that, compared to other advanced change detection methods, MCACD performs exceptionally well on our custom-built forest and grassland change detection dataset. It achieves the best F1 score and IoU values, demonstrating significant potential for practical applications in forest and grassland change detection.