the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 10 Mar 2025
Increasing flood risk in the Indian Ganga Basin: A perspective from the night-time lights
Abstract. The changing climate, intense rainfall, and geomorphological conditions within the Ganga Basin have led to recurring flooding within the area in the recent past causing severe loss of life and property. The occurrence of such flooding events has increased the need to understand the complex interplay between flood hazards, exposure, vulnerability, and risk. This study delves into flood risk within India's Ganga Basin, focusing on the flood-inducing factors, vulnerability, and exposure through the application of the Analytical Hierarchy Process (AHP) which is a Multi-Criteria Decision Making (MCDM) model. The novelty of the work is using NASA's Black Marble Nighttime Lights as a proxy for human presence and economic activities as an alternative to conventional parameters for flood exposure such as population count, and household density. The study aims to capture the dynamic nature of flood risk, driven by hydro-geomorphic controls, expanding human activities and population growth, and variations in flood resilience. We show that there is a significant increase in flood risk trend in the eastern part of the basin, particularly areas in Bihar, eastern Madhya Pradesh, eastern Uttar Pradesh and the northern part of West Bengal, identifying high flood risk zones at the pixel or cell level. The novelty of the work lies in using night-time lights as a proxy for exposure within the basin, unlike the conventional population data. This study leverages the temporal availability of the data, enabling a real-time distribution of human activities at a large scale and with greater temporal resolution.
The accuracy of the flood risk maps is validated using the historical flood-impacted data from the EM-DAT and GDIS databases, showing a satisfactory model accuracy of approximately 70 %. The findings emphasise the role of increasing human exposure and changes in rainfall patterns as the key drivers for increasing flood risk over time. This research has significant implications for flood management, offering insights for developing risk mitigation strategies that transcend administrative boundaries by identifying areas of escalating flood risk.
- Preprint
(3571 KB) - Metadata XML
-
Supplement
(1124 KB) - BibTeX
- EndNote
Status: open (until 21 Apr 2025)
-
RC1: 'Comment on egusphere-2024-3901', Anonymous Referee #1, 03 Apr 2025
reply
Title: Increasing flood risk in the Indian Ganga Basin: A perspective from the night-time lights
This study evaluates flood risk in the Ganga River Basin using the Analytical Hierarchy Process (AHP) approach, considering risk as a function of hazard, exposure, and vulnerability. A key aspect of the study is the use of night-time lights as a proxy for flood exposure, which is presented as a novel contribution. However, I find the study lacking in terms of scientific innovation, methodological advancements, and practical applications. My primary concern is the reliance on proxy variables for flood hazard assessment rather than actual hazard data. Detailed comments are as follows:
- Lack of novelty: AHP is a widely used method in flood risk assessments across various regions, including India. A simple literature search reveals numerous similar studies applying AHP in flood risk analysis (see examples below). The present study primarily replicates an established approach with minor variations in proxy variables, offering limited scientific advancement:
https://doi.org/10.1007/s11600-018-0233-z
https://doi.org/10.1007/s10661-022-10111-x
https://doi.org/10.1007/s11069-018-3392-y
https://doi.org/10.1007/s11069-019-03737-7
https://doi.org/10.5194/hess-21-2219-2017
https://doi.org/10.1186/s40677-016-0044-y
https://doi.org/10.1007/s12524-008-0034-y
- Use of proxy variables for hazard assessment: The study estimates flood hazard using proxy variables derived from DEM, rainfall, and geomorphological data, rather than employing actual flood hazard data. More robust approaches, such as using satellite-based flood observations (e.g., Sentinel data) or hydrodynamic flood models, would provide a more accurate representation of flood hazard. Recent studies have successfully integrated observed flood data into multi-criteria decision-making (MCDM) models.
- Methodological limitations and justification: The traditional AHP model relies heavily on expert judgment, which introduces uncertainty. Recent studies have addressed this limitation by incorporating hybrid deep learning models and fuzzy AHP approaches, allowing for the integration of binary flood hazard data (e.g., flood-prone vs. non-flood-prone zones) into MCDM frameworks. Furthermore, quantitative validation and uncertainty analysis are essential to ensure confidence in results. The manuscript lacks a clear justification for the chosen methodology and does not address these recent methodological advancements.
Citation: https://doi.org/10.5194/egusphere-2024-3901-RC1 -
RC2: 'Comment on egusphere-2024-3901', Anonymous Referee #2, 11 Apr 2025
reply
The manuscript “Increasing Flood Risk in the Indian Ganga Basin: A Perspective from the Night-time Lights” tackles an important issue—the escalating flood risk in the Ganga Basin—by proposing a methodology that integrates multiple geospatial datasets with NASA’s Black Marble night-time lights as a proxy for human exposure using an AHP approach. While the approach and dataset integration are acceptable in principle, several conceptual and methodological issues must be addressed before the paper is suitable for publication.
Major Comments
- Modeling Framework and Terminology
- Clarification of Model Types:
The manuscript currently confuses distinct modelling approaches. The authors refer to “physical,” “numerical,” and “hydrodynamic” models in ways that are inconsistent with established definitions. For example, physical models should be recognized as scaled, laboratory-based representations (e.g., flume models), while numerical models involve solving equations computationally. Hydrodynamic models, in contrast, specifically address the full dynamic wave Saint Venant equations in 1 or 2 dimensions derived from the Navier–Stokes framework in 3 dimensions. I strongly recommend that the authors clearly distinguish these models and ensure that each description is both scientifically accurate and well-referenced. - Terminology Consistency:
The terms “flood risk,” “flood susceptibility,” and “flood impacts” appear to be used interchangeably without precise definitions. This lack of clarity undermines the overall conceptual framework. I urge the authors to define these terms explicitly at the beginning of the manuscript and maintain consistent usage throughout the text. - Scientific Accuracy and Data Interpretation
- Rainfall Versus Precipitation:
The manuscript contains statements suggesting that “rainfall” and “precipitation” are synonymous. However, precipitation is a broader term that includes snow, ice, and hail, while rainfall does not. This distinction is critical in the context of flood risk assessment, particularly in regions where non-rainfall events may contribute to flooding. The authors should either revise this for accuracy or provide a strong justification for their interchangeable usage. - Justification of Dataset Choices:
The choice of datasets, particularly the ASTER-GDEM and CHIRPS data, is a subject of concern. Several studies indicate that higher-resolution DEMs (e.g., those derived from CopernicusDEM) might offer more reliable vertical accuracy. Similarly, while CHIRPS is a robust dataset, some recent literature suggests that GPM-IMERG may provide superior performance in monsoon-dominated regions. The manuscript would benefit from a more detailed discussion on the selection of these datasets and any limitations they might introduce. I am also really confused by the choice of the annual nighttime lights dataset as a proxy, even when the NASA Black Marble product includes daily products (https://doi.org/10.1016/j.rse.2018.03.017) which would correspond much better to flood recovery processes (see here: https://doi.org/10.1016/j.rse.2025.114645). Would the yearly averaged product not smooth out the impacts in most areas defeating the purpose or the proxy? Or are the authors claiming only to examine floods where no recovery was possible within a year to be able to see this impact on an annual scale? In any case, this is the main novelty of the paper apparently so I strongly recommend better defending this methodological choice and how this may influence their conclusions. - Methodological Presentation
- AHP Framework:
The use of the Analytical Hierarchy Process (AHP) for integrating multiple flood risk factors is not novel even though it is widely used and acceptable in this case. However, the manuscript would benefit from clearer descriptions of how the pairwise comparisons were conducted, how consistency was ensured, and how the resulting weights were validated. In some instances, the paper appears to overcomplicate the presentation of these steps. A more concise explanation, supported by relevant references, would improve readability. - Exposure and Vulnerability:
While the use of night-time lights as a proxy for human exposure is interesting, the manuscript does not sufficiently differentiate exposure from vulnerability. In several sections, vulnerability data (e.g., district-level vulnerability indices) are used in contexts that suggest they represent exposure. A thorough re-examination and clear delineation of these concepts are needed.
Minor Comments
- Figure and Visual Clarity:
Some figures, such as the workflow diagram and flood hazard maps, are not sufficiently legible at 100% zoom. I recommend that the authors provide higher-resolution images or ensure that all text and symbols are easily readable. - Scale and Resolution Issues:
There are questions regarding the spatial and temporal resolutions used, the rainfall data are daily, aggregated to monthly scales – the aggregation method is not specified – and then kept just for the monsoon months but then is compared to annual scale night-time lights data, which does not really make sense from my point of view. - References:
Several assertions in the manuscript lack adequate citation. For example, claims about the effectiveness of different modelling approaches and the influence of rainfall intensity on flooding should be supported by additional literature.
In its current form, the manuscript has several conceptual ambiguities and methodological inconsistencies that must be resolved before publication. Addressing these concerns will substantially strengthen the manuscript and enhance its contribution to the field of flood risk assessment.
-
RC3: 'Comment on egusphere-2024-3901', Anonymous Referee #3, 14 Apr 2025
reply
Title: Increasing flood risk in the Indian Ganga Basin: A perspective from the night-time lights
This study assessed the flood hazard, exposure, vulnerability and risk level of Indian Ganga Basin based on a multi-criteria risk assessment methodology (hierarchical analysis) using selected hazard, exposure and vulnerability assessment indicators. The novelty of the study lies in using night-time lights as a proxy for exposure within the basin, unlike the conventional population data.
I have a few comments/questions:
- The abstract section of the manuscript is illogical and lengthy. For example, the innovation of the paper is highlighted twice in the beginning and in the end of the abstract. It is recommended that the authors rewrite the abstract section according to research background, methods and content, results, validation, innovativeness and application value.
- The confusing use of professional terms such as “flood risk” and “flood susceptibility” means that the professional level of research manuscripts needs to be improved.
- I disagree with the discussion of the superiority of the methodology of this study in lines 652-655 of the manuscript. Accurate flood risk assessment results are obtained by driving a hydrological-hydraulic model to capture flood hazard, and then combining exposure, vulnerability, and level of prevention and mitigation. The use of multi-criteria methods to assess flood risk levels in this study may be limited by data and technical expertise. Hydrological and hydrodynamic models can simulate the depth and range of flooding and obtain more accurate risk assessment results.
Lines 652-655: “Our flood risk map of the Ganges basin was developed using an integrated hydrogeological approach and AHP methodology, which is superior to traditional hydrological and hydraulic modelling as it combines physical (geomorphological) criteria with hydrometeorological data. This emphasises a process-based understanding that overcomes the need for intensive hydrological data required for flood hydraulic models (Mishra and Sinha, 2020)”.
- The vulnerability assessment in the study used existing vulnerability products, the reliability of which was not verified. In addition, which indicators were considered in the vulnerability assessment were not clearly given.
- The novelty of the work lies in using night-time lights as a proxy for exposure within the basin, unlike the conventional population data. However, in the flood risk assessment framework based on multi-criteria methods, using night light data instead of population and economic data as an innovation in flood risk assessment seems to be somewhat insufficient for the entire study, but it is acceptable.
In flood risk assessment, research innovation would be more prominent if hazard assessment used flood inundation information obtained based on remote sensing data while taking into account changes in flood vulnerability.
- The dynamic trends of flood risk levels should be placed in the results section of the study rather than the discussion section, which should focus on the superiority of using nighttime light data for flood exposure and risk assessment.
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 111 | 40 | 7 | 158 | 13 | 4 | 4 |
- HTML: 111
- PDF: 40
- XML: 7
- Total: 158
- Supplement: 13
- BibTeX: 4
- EndNote: 4
Viewed (geographical distribution)
| Country | # | Views | % |
|---|---|---|---|
| United States of America | 1 | 58 | 35 |
| United Kingdom | 2 | 22 | 13 |
| China | 3 | 18 | 10 |
| India | 4 | 17 | 10 |
| undefined | 5 | 11 | 6 |
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
- 58