Good points and thank you for these constructive comments. We will address them.

We sincerely thank the reviewer for the careful reading of our manuscript and for the constructive and helpful comments. We are pleased that the reviewer found the manuscript generally well written and recognized the regional relevance and practical value of the harmonized dam and reservoir dataset. We also appreciate the specific suggestions regarding the need to provide more technical details on data harmonization, duplicate removal, dam–reservoir matching, manual digitization, temporal mismatches, and the non-independence of some source datasets.

We agree that improving the transparency and reproducibility of these methodological steps will strengthen both the manuscript and the dataset. In the revised version, we will provide additional details in the Methods and Discussion sections and add a dedicated limitations subsection. Please find below our preliminary responses to each comment.

1. Abstract: Please define “small-scale infrastructure” and provide the corresponding area, storage-capacity, or dam-height threshold.

In our manuscript, “small-scale infrastructure” refers to minor and small reservoirs of the definitions proposed by FAO (FAO, 2001), which ranges from 0.010 to 9.99 km² surface area. To avoid any ambiguity, we will revise the manuscript to strictly follow the FAO classification, as shown below:

Reservoir category             Surface area range at USL (km²)

Major    ≥ 1,000

Large     100–999

Medium 10.0–99.9

Small     1.00–9.99

Minor    0.010–0.999

1. Introduction: The terms “large” and “small” dams and reservoirs should be precisely defined throughout the manuscript. If the classification is based on reservoir area, please provide the corresponding area ranges.

In the revised manuscript, we will clearly define these categories based on the definitions proposed by FAO (FAO, 2001).

1. Line 76: The research gap is described as the “underestimation” of West African dams and reservoirs in existing datasets. Does this refer to the number of dams, reservoir area, storage capacity, positional accuracy, attribute completeness, or a combination of these factors? Please define the specific dimensions of underestimation more clearly.

In the current version, the term “underestimation” refers to the number of dams and reservoirs reported in existing datasets. We will address this in the revised version of the manuscript through and improved definition or a change in terminology.

1. Please explain how duplicate records were identified and removed, including the matching criteria, thresholds, and whether the process was automated or manual.

The harmonization was performed manually in QGIS after merging the selected source datasets. Separate harmonized attribute tables were created for dam points and reservoir polygons. Duplicate records were identified using a combination of spatial and attribute-based checks. First, records with identical or nearly identical geographic coordinates were flagged as potential duplicates. Second, overlapping or very closely located features were visually inspected in QGIS. Third, available attributes such as dam or reservoir name, country, river or basin name, reservoir area, storage capacity, construction year, and source dataset were compared to confirm whether the records referred to the same infrastructure.

When duplicate records were confirmed, only one representative record was retained in the compiled dataset. Attribute information from the duplicate sources was not simply discarded; rather, consistent information was retained and used to improve the completeness of the final record. In cases where attributes differed across sources, priority was given to the most complete and plausible record. Ambiguous cases were checked manually before final inclusion.

We will revise the method section accordingly.

1. Please clarify whether dam points and reservoir polygons have a one-to-one relationship in the compiled dataset. How were dam points linked to reservoir polygons?

Dam points and reservoir polygons do not always have a strict one-to-one relationship in the compiled dataset. This is because the selected source datasets differ in their feature types and purposes. Some datasets provide dam locations as points only, while others provide reservoir polygons without associated dam points. In addition, some large or complex reservoir systems may include more than one dam-related point, while some small dams may not have a clearly mapped reservoir polygon in the available products.

Where possible, dam points were linked to reservoir polygons using spatial overlay and proximity analysis in QGIS. A dam point was associated with a reservoir polygon when it was located within, on the boundary of, or immediately downstream of the polygon. This spatial matching was supported by attribute information where available, including dam or reservoir name, country, river or basin, reservoir area, and source dataset. Ambiguous cases were manually inspected using the spatial layers and, where necessary, satellite imagery or base maps.

We will clarify this in the revised manuscript.

1. Please clarify whether each record retains its original data source and indicates any modifications made during harmonization.

Thank you for this important comment. We agree that data provenance is essential for the transparency and reuse of the harmonized dataset.

In the current version of the compiled dataset, each dam or reservoir record does not retain the full list of all original datasets in which the same infrastructure may have appeared. During harmonization, when duplicate or overlapping records were identified, we retained one representative record. The source indicated for each final entry corresponds to the dataset that provided the entry itself and/or the majority of the retained attribute information. Therefore, the source field should be interpreted as the main contributing source for the harmonized record, rather than as a complete record-level provenance history across all overlapping datasets.

However, all original datasets used to build the harmonized database are cited throughout the manuscript and are also listed in the README file associated with the published dataset. In the revised manuscript, we will clarify this point to avoid any misunderstanding. We will also explain that the main harmonization steps included standardizing attribute names, harmonizing units where needed, removing duplicate entries, and retaining the most complete and consistent information when the same dam or reservoir was reported by more than one source.

1. Please report the sample size and selection criteria for the surveyed dams and villages in the Methods and clarify whether the survey covered all known dams in the watershed.

We will revise the method section and provide the information that the survey covered all known dams in the area.

1. The field survey was conducted in 2024, whereas some source datasets are much older. Please consider temporal mismatches when interpreting missing dams and assessing dataset accuracy.

We agree that temporal mismatch is a key source of uncertainty. We will consider this in the revised manuscript when discussing missing dams and assessing the accuracy of the dataset.

1. Figure 4: Please define the meaning of “lake control.

The term “lake control” comes from the HydroLAKES database. It refers to a natural lake that is influenced or regulated by a control structure, such as a dam or outlet regulation structure. We will define this term in the caption of Figure 4 and/or in the relevant section of the text in the revised manuscript.

1. Many of the selected datasets overlap or inherit information from other datasets. For example, some newer products incorporate records from GRanD, GLWD, HydroLAKES. Did the harmonization procedure account for this lack of independence?

We agree that the selected datasets cannot all be considered fully independent. Several dam and reservoir products share common source information or inherit records from earlier datasets such as GRanD, GLWD, and HydroLAKES.

Our harmonization procedure accounted for this issue mainly through duplicate identification and data-source tracking. When the same dam or reservoir appeared in multiple datasets, it was treated as a duplicate record rather than as independent evidence of multiple infrastructures (dams and reservoirs). The final compiled dataset therefore aimed to retain one harmonized record per identified infrastructure while preserving source information for traceability.

We will discuss this more explicitly in the revised version.

1. Table 4: The current comparison of attribute availability is rather coarse. Please consider listing the individual attributes provided by each dataset; this should not require excessive additional space and would substantially improve the usefulness of the comparison.

Thank you for this useful suggestion. The existing datasets contain many individual attributes. For example, GRanD alone includes more than 45 attributes. Listing all attributes from each dataset in Table 4 would therefore make the table too large and difficult to read in the main manuscript. For this reason, we grouped the available attributes into four main categories and presented these categories in Table 4 to provide a concise overview of attribute availability across datasets.

1. Lines 307–311: 148,671 MCM represents approximately 52.5%, rather than 55.61%, of 283,032 MCM.

Thank you. We will correct the percentage in the revised manuscript and check all numbers.

1. Lines 399–400: If the compiled dataset contains 32% of the dams observed during the field campaign, the gap should be approximately 68%, not 88%.

We will correct the statement in the revised manuscript.

1. The field-based assessment was conducted in only one watershed. Please discuss more carefully whether its results can be generalized to the whole of West Africa.

The Upper Bandama watershed provides a valuable benchmark because it contains many small reservoirs and represents a data-scarce context where existing datasets often have limited coverage. We fully agree that this is only one watershed and does not capture the full diversity of environmental, hydrological, institutional, and data-availability conditions across West Africa, but it is currently the only validation data available to us for this assessment and thus of high value. We will address this in the revised discussion.

1. I recommend adding a limitations subsection in discussion. This section should address the use of only one watershed for field evaluation, the non-independence of source datasets, and potential uncertainty in the field-reference data itself.

We will add a dedicated limitations subsection to the discussion.