The paper is quite boastful and ignores the complexities of projecting VIIRS global nighttime light back to 1992.   Earth Observation Group provides access to the DMSP monthly nighttime lights from 1992-2021 and VIIRS from 2012-2025.  EOG's offerings are clearly the first open-access time series of monthly nighttime lights spanning 1992-2025.  DMSP and VIIRS nighttime lights differ from each other in several key ways: A) VIIRS DNB pixel footprints are 42+ times smaller than DMSP. Many small lights detected by VIIRS are absent in DMSP. B) VIIRS has lower detection limits and a wider dynamic range.  In contrast, DMSP nighttime light observations use 6-bit quantization and frequently saturate in bright city centers. C) The VIIRS overpass time is typically between midnight and 03:00 local time.  The early DMSP record (1992-2013) had mid-evening overpass times, between 19:30 and 21:30. The DMSP extension series (2013-2021) has pre-dawn overpass times.  Nighttime lights have variable diurnal patterns.  This paper's data from 1992 to 2011 are speculative and cannot be recommended for quantitative use.  The earliest VIIRS data are from 2012.  The paper's title is thus misleading. The paper takes an ill-informed and swaggering approach to generating monthly nighttime lights from 1992-2011.

This study addresses a crucial demand in nighttime light (NTL) remote sensing research by developing a long-term, high-resolution monthly VIIRS-like NTL dataset (MVNL) for 1992–2024, and the research work is of great scientific value and practical significance. Fusing multi-source NTL data to extend the temporal coverage of high-resolution products is an inevitable and essential approach to studying long-term urbanization and socioeconomic dynamics over the past decades, despite inherent limitations from the spatiotemporal inconsistencies of original DMSP-OLS and NPP-VIIRS data. This study makes prominent contributions to the field: it constructs the first 500 m-resolution monthly VIIRS-like NTL dataset spanning 1992–2024, filling the gap of high-resolution (500 m) multi-source fused long-term NTL products (only 1 km-resolution products are currently available); the reconstructed dataset achieves high consistency with existing EOG products (R²=0.96 at city scale for overlapping months in 2012, R²=0.98 at city scale for 2012 annual composite), and is further verified against radiance-calibrated DMSP products (R²>0.94 at city scale), demonstrating high reliability for city-level applications; additionally, the dataset has been freely shared online, providing a valuable and accessible data resource for global urbanization, socioeconomic analysis and related cross-disciplinary research. Overall, the research design is rigorous, the technical route is feasible, and the data product has important application prospects. I recommend the manuscript for minor revision and subsequent acceptance. Specific revision suggestions are as follows:

1. Since there are already existing attempts at 1 km-resolution long-term NTL products, the title "The World's First Long-Term Global 500 m-Resolution Monthly VIIRS Nighttime Lights Dataset (1992–2024)" may have certain ambiguities. It is necessary to clearly specify whether the dataset is the world’s first 500 m-resolution product, the first global-scale product, or the first monthly-scale product, so as to avoid misunderstandings by readers.
2. The DMSP data covers 1992 to 2013, while the NPP-VIIRS data covers 2012 to the present, with the overlapping years of the two datasets being 2012 and 2013. It is advisable to clearly state these overlapping years in the abstract, and explicitly explain that the 2012 NPP-VIIRS data was used as the validation dataset in the study design. Otherwise, readers may be confused about why the validation accuracy is based on 2012 data.
3. The abstract mentions using 2013–2024 VIIRS data as a reference to reconstruct 1992–2013 DMSP data. However, from the perspective of the technical route, the reconstruction actually mainly uses the 2013 VIIRS data (the overlapping year) to establish the model. It is necessary to more clearly describe how the 2013–2024 VIIRS data is used as a reference for DMSP data reconstruction. For example, if my understanding is correct, the 2013 data is mainly used to establish the NightNet model, while the 2013–2024 time series data is mainly used to correct the temporal variation trend of DMSP data.
4. Supplement the detailed methodological principles of the joint intra-annual and inter-annual interpolation strategy mentioned in Line 205 (used for gap-filling the 1992–2012 monthly DMSP series), as the current manuscript only mentions the application of the strategy but lacks a clear explanation of its core principles, which affects the reproducibility of the research.
5. For the complex algorithms involved in the spatiotemporal fusion model constructed in this study, clearly distinguish existing mature algorithms/models and original algorithms of this study. For the adopted existing algorithms, add relevant reference citations and briefly explain the reasons for selection (e.g., core advantages, wide and successful application in remote sensing data spatiotemporal fusion); for the use of existing programs, provide a brief description as well.
6. Optimize the visualization effect of Figure 8: the global NTL distribution pattern is unclear, it is recommended to add local enlarged views or adjust the color matching to enhance the readability of spatial distribution characteristics; in addition, supplement the spatial distribution maps of the original DMSP and NPP-VIIRS data (especially local enlarged views) for comparison, which can help readers intuitively identify the differences between the reconstructed product and the original multi-source data.
7. It is suggested (not mandatory) to add a pixel-scale accuracy comparison across different regions in the precision validation part. NTL data quality and precision are inherently different in different regions (e.g., DMSP saturation leads to lower precision in highly urbanized areas, and precision differences may also exist across latitudes/geographical regions). Pixel-scale regional precision assessment can help users apply this 500 m high-resolution product more scientifically and objectively, and better reflect the advantages of the high-resolution product proposed in this study.
8. If there are other existing similar NTL products, adding a comparative analysis with these products can further highlight the advantages of this study’s product. Relatively speaking, for manuscripts focusing on data product development, the analysis of the correlation between NTL and population, economic activities, energy consumption, etc., may not be so important (if the manuscript has page limitations).

This paper presents a valuable and well-constructed dataset that bridges DMSP-OLS and NPP-VIIRS observations to produce a long-term, monthly, and consistent nighttime light (NTL) product. The effort to reconstruct VIIRS-like monthly data from historical DMSP records is important, and I believe this dataset will be highly useful for a wide range of socioeconomic and urban studies. The manuscript is generally well written, and the results are promising. I have several minor suggestions that could further strengthen the paper:

1. The manuscript mentions the saturation and blooming effects in DMSP data. It would be helpful to provide quantitative evidence demonstrating how well saturation and blooming effects are mitigated compared to the raw DMSP data.
2. Please ensure consistent use of the term “nighttime light (NTL)” throughout the manuscript (e.g., around Line 310).
3. The discussion of seasonal variation (e.g., Figures 19 and 20) is informative, but could be strengthened by including quantitative metrics to more clearly demonstrate how well seasonal dynamics are captured. In addition, a comparison with the seasonal patterns derived from monthly DMSP data would be valuable, as it could further highlight the improvements and added value of the proposed dataset.
4. The model is trained using data from 2013, while overlapping data are available for both 2012 and 2013. Please clarify why 2012 was not included. A comparison between models trained on different years (e.g., 2012 vs. 2013) could also help demonstrate the robustness of the approach.

1. The proposed network consists of three major functional modules; the overall architecture appears relatively complex. While the design is technically sound, it would be helpful to include, in Section 3.2, a discussion comparing the proposed model with simpler baseline networks (e.g., conventional CNN-based architectures). Such comparisons would help clarify the rationale for adopting the current design and its added value.