Comments

This paper downscales national GDP estimates across a global grid of 30 x 30 arcsec pixels. This is an interesting objective and recognizes that little work has indeed been done to move beyond people-based GDP-scaling to one that also considers the distribution of economic activities. The methodology as well as several involved assumptions and uncertainties are described transparently.

However, I have several concerns regarding this paper and the quality of the dataset. In addition to the detailed comments provided below, overall, it appears that the paper attempts to integrate two papers rather than producing a single focused paper; the paper namely both documents the creation of a global GDP map, and attributes much of the paper’s attention (see e.g. the discussion section) on Thailand and Thai-specific issues.

1. The paper augments established European Commission data to differentiate global land use by residential, non-residential, and cropland uses. However, it is assumed (p.6) that residential use (“RES”) represents the service sector of the economy. That is a very rough proxy given that this includes the housing of those who work in non-residential areas (the “industrial sector”), as most people do. Moreover, the non-residential areas being classified as the ‘industrial’ sector, if I understand the classification scheme correctly, pools together any services and manufacturing and other sectors as ‘industrial,’ separately from ‘services’. This appears to be inappropriate and thus call into question whether the global map is able to distinguish between sectors. The data do appear to possibly reasonably allow for a global GDP map, without sectoral differentiation, that downscales national GDP estimates given local non-residential land use.

1. Claims such as “in the United States, industrial GDP is widely dispersed regardless of urban areas” are interesting but also bold, given that the observation comes from the East coast of the USA which is relatively agglomerated (how are “cities” defined in the paper?) and paired with serious uncertainty, given that the validation of the global dataset is done for Thailand but not for the rest of the world. Ideally, analytical claims should be made only for regions for which the data are also validated to not over-assert the validity of the data that underpin the insights. In any case the validity of the findings could be asserted more carefully. It would also be helpful to compare the insights against to standing knowledge, whether from estimates in other papers or also reports (e.g., such as the 2012 ‘Urban America’ McKinsey report).

1. The paper could do more to underpin assumptions with field knowledge, in particular on how the assumptions could drive the outcomes observed in the global map. For instance, on p.6 it is stated that “the service GDP was distributedonly in pixels within cities and the amount of distributed GDP was proportional to the population density of the citywhere the pixel is located”. This appears to in effect assume away any service sector presence outside of urban areas, which is unrealistic, and that the amount of GDP attributed to a pixel is contingent on city density —other than the size of the city— which indeed drives productivity but not overall output levels as those instead respond predominantly to city scale.

Further comments

- The narrative flow and grammar should be checked closely throughout the manuscript (see, e.g., the first five sentences of the abstract).

This manuscript develops a sectoral GDP map (for service, industry, and agriculture) at 30 arcsec resolution and explores its application in disaster risk analysis. The authors generate land-use data and polulation data to downscale national-level GDP  to derive spatial distribution results. By providing high-resolution global sectoral GDP maps, this dataset offers more detailed geospatial information to support disaster risk analysis and economic loss assessments.

The methodology and limitations in the manuscript are clearly discussed. However, the validation and analysis of the data itself need to be strengthened. Additionally, the Discussion section should be reconsidered in terms of its length and content.



Specific Comments:

1.  The Introduction section should include a discussion of other existing GDP spatial datasets, covering their methodologies, spatial resolutions, and the challenges in existing GDP mapping processes.

2. The study assumes that service-sector GDP is primarily distributed in high-population-density areas, but certain economic activities—such as high-end financial services and tourism—do not necessarily follow this pattern. For example, the financial district in Manhattan has an extremely high GDP density despite relatively low residential population density. Have the authors considered such spatial distribution patterns of economic activities?

3. Why did the authors choose the GRUMP dataset to account for urban effects instead of other datasets? A brief explanation for this choice would strengthen the methodology section.

4. The validation was conducted in only seven regions of Thailand, but Thailand’s economic structure may not be representative at a global scale. For example, Western economies are more dependent on the service sector, while industrial and agricultural distributions vary significantly across different regions. Have the authors considered additional validation in countries with different economic structures, such as the United States, China, or Germany?

5. A comparison with other existing GDP products or remote sensing proxies should be included to better highlight this dataset’s advantages.

6. Since the study aims to provide a globally applicable dataset, the Thailand case study in Section 4.1 should be presented as a supporting example rather than the main focus. It is recommended that the authors strengthen the discussion of the dataset itself, particularly regarding accuracy assessment, comparisons with existing datasets, spatial details, and temporal variation analysis. Additionally, by reducing the emphasis on the Thailand case study and discussing broader disaster analysis applications, the authors can better highlight the dataset's global applicability.