General comments

The paper applies the main concepts of the SMART approach to a catchment in the Himalayas, highlighting the relevant role of community participation in addressing the limitations of data-scarce regions and observation-based warning thresholds. To underline the importance of increasing population preparedness in catchments presenting short response times and urban settlements along rivers, the variability of recent hydrological events is investigated. The paper is generally well-organised, and the issue of improving the resilience against floods in catchments characterised by flash floods is worth investigation.

Specific comments

1. Figure 1: The shape of the watershed does not correspond with the purple one in the inset. Please check;
2. Lines 140-144: The authors should include an image to help the reader understand the location of the mentioned stretches;
3. As community engagement is an essential part of the work, I encourage the authors to expand the description of the PRA exercises, further clarifying how their results could be applied within a real-time early warning system. Since EWS typically rely on thresholds based on observed hydrological variables (e.g. water levels), how can community information be incorporated into such chains for prediction purposes? Moreover, it would be interesting to compare the information provided by the participants with that derived from recorded hydrological observations for some events, e.g. by representing on the same map the locations where the most critical flood issues were recorded and remembered. This will further clarify the role of community memories. Finally, since flood perception is often influenced by past flood experiences, it would be interesting to understand if the authors noticed differences in the information provided by the participants according to their age;
4. Line 234: In addition to what was mentioned at the previous point, please clarify the criterion adopted to join the “community engagement” with the flood alert thresholds reported in Table 2.

Technical corrections

1. Page 1, line 35: In addition to the acronym, please define SMART;
2. Please, make sure that the reference section actually includes all the cited papers: several works mentioned in the paper are actually missing, such as Papalexioux & Montanari, 2019 (line 42), Rentschler et al., 2022 (line 46), Gu et al., 2019 (line 60), and many others;
3. Line 49, a point is missing before “The”;
4. Line 57, a point is missing before “Rapid”;
5. Line 62: Please check the consistency of “reach become”;
6. Lines 180-181: R1, R2, R3, and R4 are actually RG1, RG2, RG3, and RG4. Please correct;
7. Line 383: Please check the sentence;
8. Reference section: Some references are listed twice. Please check.

Review

The study presents an integrated approach to the design of early warning systems for flash floods in both urban and mountainous environments, combining real-time monitoring technologies with the active involvement of local communities.

The authors implement a high-resolution hydrometeorological network, based on LiDAR sensors and advanced measurement instruments, aimed at providing a detailed characterization of the rainfall and hydrological regime of the study basin.

The data collection and analysis highlight a marked spatial variability in precipitation (up to more than 180 mm between stations only a few kilometers apart), a crucial factor for forecasting localized flood events.

The SMART model constitutes the conceptual core of the study and, although not clearly defined, is described as a dynamic and adaptive system. The comparison with global reanalysis (ERA5) and satellite (GPM) datasets shows that these sources fail to adequately capture the complexity of precipitation patterns in mountainous areas. In contrast, the SMART approach proposed here, based on real-time local data, integrates basin dynamics and adapts alert thresholds in a context-specific manner.

A key methodological feature is the definition of thresholds based on percentiles of water level data, validated through community participation, which makes the system more flexible and responsive to actual environmental conditions.

Although the work represents an original contribution to the literature on early warning systems for extreme events, it also presents some limitations, several of which are acknowledged by the authors themselves.

General Concerns

1.    Limited spatial and temporal extent – The study focuses on a single small catchment (Bindal, 44.4 km²) and a relatively short observation period of approximately one year (September 2022 – August 2023). This timeframe is insufficient to robustly assess the system’s performance with respect to interannual variability or its ability to capture rare extreme events. The authors themselves acknowledge that the hydrological response of the basin is strongly influenced by local-scale factors, underscoring the need for further testing and validation across broader spatial and temporal scales.

2.    Lack of operational validation – The study does not include a verification of the system under real operational conditions, nor does it present application-based simulations for future events. Moreover, no performance metrics are provided — such as lead time, false alarm rate, or the accuracy of dynamic thresholds — which are essential for a quantitative assessment of the system’s effectiveness.

3.     Dependence on percentile thresholds – The system relies on statistical thresholds derived from percentiles of water level data; however, the study does not provide an in-depth analysis of the model’s sensitivity to variations in these thresholds, nor does it address its ability to manage exceptional or out-of-scale events.

4.    Non-formalized community involvement – Although the study places significant emphasis on local community participation, it does not present a structured and replicable methodology for systematically integrating community knowledge into the decision-making process. Furthermore, operational details on how surveys and consultations were conducted are lacking. The overall effectiveness of the system largely depends on the level of community engagement and technical capacity — factors that may limit its transferability to other socio-cultural contexts.

5.    Absence of predictive hydrological or hydraulic modelling – The study focuses primarily on monitoring activities and descriptive data analysis but does not incorporate physical or predictive runoff models capable of simulating future scenarios or assessing the impacts of anthropogenic changes and climatic variations. The authors themselves acknowledge the need to integrate hydrological modelling components to enhance the operational effectiveness of the early warning system.

In summary, the study is characterized by methodological innovation and strong practical potential, but it requires further validation, the development of predictive components, and operational testing in order to strengthen its effectiveness and enhance both its applicability and transferability on a broader scale and to other catchments.

Specific Comments

•    Abstract: It should be clarified more explicitly whether the implemented procedure is intended for nowcasting or forecasting purposes. Although the term forecasting is mentioned in line 101 of the introduction, the use of real-time precipitation and runoff monitoring tools might suggest a nowcasting application, which is generally impractical for a catchment of such limited size. Therefore, the operational objective of the system should be specified more clearly. It is presumed that the real-time data were primarily used to calibrate a forecasting model.

•    Figure 2: It would be advisable to clarify whether the four key components hold the same level of importance within the methodological framework. The graphical representation appears to suggest equal weighting, but a brief explanation in the text would help to better understand any hierarchical or functional relationships among them.

•    Lines 140–145: Including a geographic reference figure in this section would enhance the spatial understanding of the study area and clarify the subdivision of the river segments. At present, Figure 3 provides only partial information, and its placement far from the relevant text reduces its immediacy and readability.

•    Paragraph 3.1 – Community Interaction: In this section, it is not clearly explained how community interaction contributed to the definition or adjustment of the thresholds. Including examples of the questions asked to participants, along with a map of the most vulnerable areas and an explanation of how these maps were produced, would make the methodology more transparent and easier for the reader to understand.

•    Lines 175–180: It is recommended to revise Figure 1 by moving part 1a to this section, in order to immediately show the location of the sensors and facilitate comprehension, avoiding the need for the reader to flip back several pages. It would also be useful to indicate the distance of the discharge measurement sensors along the river channel.

•    Lines 188–193: It may be more effective to place the entire Figure 1 in this section to ensure consistency between the text and the illustrations. Alternatively, a new figure could be added in Chapter 2, specifically dedicated to the description of the study area.

•    Line 240 (Table 2): It is advisable to adjust the table background, as the first row is difficult to read due to the low contrast between the text and the background color. Increasing the contrast would significantly improve readability.

→ Some comments from the interactive discussion are not repeated here, but I fully agree with them and suggest the authors to address those points as well.