This manuscript presents a 30 m resolution dataset of paddy rice spatial distribution and cropping intensity across South and mainland Southeast Asia for the years 1995, 2005, 2015, and 2024, derived from multi-year composites of Landsat and Sentinel-2 imagery using a refined phenology-based approach. The manuscript is generally well-structured, and the authors have compared their results with several existing rice mapping products as well as FAO statistics. However, there are several methodological and validation concerns that need to be addressed.

1. The manuscript reports validation using 23,396 samples across four decades. However, only the 2024 samples were manually interpreted, and earlier-year labels were transferred using a sample migration algorithm. While this algorithm is reasonable for constructing historical training data, the migrated samples may not be considered fully independent validation data, as they mainly represent long-term stable pixels and may not capture areas of actual land-use change, such as newly cultivated or abandoned fields. As a result, using them to assess accuracy in earlier years could overestimate algorithm performance.
2. The dataset is based on half-monthly NDVI/LSWI/EVI composites and multi-year (3-5 year) compositing to represent individual reference years. While this strategy likely improves data availability and reduces noise, it may also smooth short-duration cropping signals or interannual variability, particularly in regions with triple cropping or short-cycle varieties. Providing additional assessment of temporal aggregation effects would improve confidence in the annual representativeness of the product.
3. The manuscript introduces the “enhanced peak detection” and “false peak elimination” methods. It is suggested to include quantitative comparisons with existing phenology-based approaches.
4. Key parameters, including Whittaker smoothing (λ = 300), minimum season length (120 days), and many thresholds, are applied uniformly across all regions, years, and management regimes. Such uniform choices may fail to capture short-duration varieties or respond appropriately to differences in irrigation practices and cropping calendars. These parameters raise concerns regarding the method’s transferability across regions, years, and management conditions.
5. The manuscript constructs custom composites of Landsat and Sentinel-2 imagery instead of using the Harmonized Landsat Sentinel-2 (HLS) product. It is suggested to clarify whether this choice affects paddy rice spatial distribution and cropping intensity detection.
6. The dataset could be further enhanced by including additional confidence or uncertainty layers, considering potential downstream applications.