Review Reports - Differentiated Spatial-Temporal Flood Vulnerability and Risk Assessment in Lowland Plains in Eastern Uganda

Round 1

Reviewer 1 Report

General comment: Although it is noteworthy that authors have undertaken a model cascade to derive flood maps for a specific case study, it is not clear what is the real scientific contribution or innovation. For instance, a new (combined) procedure, an innovative tool, a discovery while working, or maybe emphasizing climate change influence on the Manafwa river system, might be recommendable in this respect. Otherwise, a technical publication is preferable, rather than a scientific one. In addition, there are other observations:

Introduction:

· Page 2, line 63: HEC-HMS, HSPF, SWAT are not models, but modelling tools (or evaluation tools), used to build models. Same on line 68 and 77.

Materials and Methods:

· Page 4, figure 1: The coordinate font size is very small, is not readable. Idem with the legend. Increase the figure resolution.

· Page 4, line 144: indicate the spatial resolution of landuse maps. Is it comparable with 30 or 25m?

· Page 6, line 206. For the catchment in study, perhaps the spatial resolution 30 x 30 m is suitable. However, I doubt it for the hydrodynamic study. Terrain changes along slopes and riverbanks might not be capture properly. I suggest reducing or refine the grid resolution (e.g. bathymetry) to 5m or less in that respect (along the river network), based on other data source (e.g. local survey or bathymetry).

· Page 6, line 229: When using 2D modelling (full Saint-Venant equations), steady flow is not applicable. Unsteady analysis is mandatory. If there is no hydrograph available, then build a synthetic one, whose maximum is the computed discharge from the HMS model. This observation is made also for line 234 (computed discharge hydrographs from steady flow?).

· In section 2.5 please indicate the spatial resolution of the LULC imagery.

· Page 9, line 309, delete the dots after equation 4.

· Page 9, figure 5, eliminate decimal places on Y axis.

· In figure 8, there is no clear difference between the 4 maps.

· Page 12, line 368: 9.0 km². Idem on the next line and further.

· Page 12, lines 378-383, why flooded areas have decreased? Is there any reason?

Author Response

Research premised on the fact that limited research exists in literature for Uganda river basins. The novelty of the present study is to combine the physically-based distributed hydrological model SWAT with the hydraulic model HEC-RAS for ﬂood prediction in Eastern Uganda, which has not been done before for the tropical catchments and for small watersheds.

Introduction:

Page 2, line 63: HEC-HMS, HSPF, SWAT are not models, but modelling tools (or evaluation tools), used to build models. Same on line 68 and 77.

Rewritten as modelling tools

Materials and Methods:

Page 4, figure 1: The coordinate font size is very small, is not readable. Idem with the legend. Increase the figure resolution.

Figure resolution has been increased

Page 4, line 144: indicate the spatial resolution of landuse maps. Is it comparable with 30 or 25m?

spatial resolution is 30 X 30 m and has been inserted

Page 6, line 206. For the catchment in study, perhaps the spatial resolution 30 x 30 m is suitable. However, I doubt it for the hydrodynamic study. Terrain changes along slopes and riverbanks might not be capture properly. I suggest reducing or refine the grid resolution (e.g. bathymetry) to 5m or less in that respect (along the river network), based on other data source (e.g. local survey or bathymetry).

The 1D model was connected to the floodplain and a 2D computational mesh was created at 100 x 100 m grid size. Although the cell size is rather large, considerable hydraulic details are still retained within a cell using the 2D Geometric Pre-processor. The algorithm preprocesses cells, and cell faces to develop detailed hydraulic property tables (elevation versus wetted perimeter, elevation versus area, roughness, etc.) based on the underlying terrain (5 x 5 m in this case).

Page 6, line 229: When using 2D modelling (full Saint-Venant equations), steady flow is not applicable. Unsteady analysis is mandatory. If there is no hydrograph available, then build a synthetic one, whose maximum is the computed discharge from the HMS model. This observation is made also for line 234 (computed discharge hydrographs from steady flow?).

Study used a combined one-dimensional and two-dimensional hydraulic (1D2D)
model

In section 2.5 please indicate the spatial resolution of the LULC imagery.

spatial resolution is 30 X 30 m and has been inserted

Page 9, line 309, delete the dots after equation 4.

dots removed

Page 9, figure 5, eliminate decimal places on Y axis.

Decimal places eliminated on the Y axis

In figure 8, there is no clear difference between the 4 maps.

The difference is indicated in the graphs adjacent to maps for different return periods.

Page 12, line 368: 9.0 km². Idem on the next line and further.

corrections made for km² throughout the text

Page 12, lines 378-383, why flooded areas have decreased? Is there any reason?

This comment is rather not clear because in the text, I explain increase in flood areas

Reviewer 2 Report

The paper presents flood risk assessment for a particular river catchment in Uganda. The authors give an extensive introduction to the problem. Then they present the methods and tools they uses which are: a hydrological model of river flow, soil and water assessment tool, ArcGIS mapping facility. They describe the studied area, the modelling setup, the carried out experiments. They analyze their results such as flood vulnerability.

The paper is well written, easy to follow and apparently deals with a problem of regional importance - the flood vulnerability along an important river.

There are very rare writing mistakes like for example:

line 76: ... then fed into the hydrodinamic ....

line 500: ... In general its concluded that ...

The paper merits publication as it is.

Author Response

line 76: ... then fed into the hydrodinamic ....

rewritten as were then fed into the hydrodynamic.....

line 500: ... In general its concluded that ...

has been rewritten as.. In general it is concluded that...

Reviewer 3 Report

Comments for author File: Comments.pdf

Author Response

1. Line 17,19,94,103, 379 : In a few places, it is HEC-RAS, and in some others, it
is mentioned as HECRAS; make it uniform.

I have replaced HEC-RAS with HECRAS throughout to ensure uniformity

2. Line 24-25: Some land use/cover classes are mentioned in lowercase and a few others in title case; make it uniform.

All land use/cover classes written in lowercase

3. Line 64: whether it is climate change land use impacts or climate change and
land use impacts?

Rectified and written as climate change and land use impacts

4. Line 158: add the full form of HRUs

Written in full- Hydrological Response Units

5. Line 23: it is as Figure.2; Line 252 is as Fig.3 maintain uniformity.

The comment was not clear to me and as such could not address it

6. The role of land use/cover change analysis on flood vulnerability has not been discussed in detail except for subsistence farmland.

The area under study is rural and the main Land use is subsistence farmland. Due to population growth there is significant conversion into farmland

7. Figure .4 shows river cross sections, stream centerlines, and flow lines, which are generally used in 1D. But in the manuscript, it seems to be that 2D has been used for deriving flood inundation mapping. Clarify.

A combined one-dimensional and two-dimensional hydraulic (1D2D)
model was used in the study

8. There are several pointless spaces in between words, such as the one on line 149 between "from 1981 to 2013." Fill in any blank spaces that are superfluous
throughout the paper.

Blank spaces have been filled

9. In Figure 1: Map of the Study area; Legends are not clearly visible.

Map replaced with a map with a better resolution

10. In Figure 6: Observed and simulated monthly streamflow hydrographs for the calibration Period- 2000-2010 and the validation period 2011-2013 (separated by the vertical dashed line) for 2008 land Cover; Legends inside the chart hides the peaks in data. Rearrange the Legend in another place.

Legend put in another place as notes beneath the caption of Figure 6

Reviewer 4 Report

In this work, the authors perform a flood-inundation application at the Manafwa river, by comparing a combination of the SWAT and HEC-RAS models along with streamflow observations. Please consider addressing some issues found below:

1) In the Introduction and the Analysis is mentioned the debate about the complexity of the hydraulic model (e.g., whether to use a1D/2D model; see also the discussion by Horritt and Bates, 2002). An overview of this complexity issue (and other issues and challenges in flood models) can be found in works such as by Merwade et al. (2008), Neal et al. (2012), Mishra et al. (2022), who most concluded that both type of models have limitations and advantages depending on the specific case.

2) Please give more information on how were the rainfall-runoff models calibrated and validated, how many parameters were included in this procedure and what were their values, were there any of the parameters linked to high variability or uncertainty (causing one of the models to under/overestimate the low or the peak flows and why), etc. These comments will also help the readers follow the answers to the questions that the authors raised, i.e., (1) how will the coupled hydrology-inundation model perform in order to attain proper flood assessment in a data-scarce area? and (2) can the developed model-based flood assessment fill the knowledge and technical gap in the development of flood protection measures and flood risk management? Please consider further explaining how the authors' research answers these questions for the area of their application.

3) The authors mention a sensitivity analysis for the estimation of the hierarchy of the input parameters; please consider giving more information on the conclusions of this sensitivity analysis (for example, Dimitriadis et al., 2016 performed a similar analysis and concluded that the roughness parameter and the rainfall had the largest variability/uncertainty among many input parameters related to hydrology, hydraulics, topography, etc.).

4) In the widely used hydrodynamic models please consider including FLO-2D (https://flo-2d.com/), which is one of the first hydraulic models to perform sediment transport (the HEC-RAS methodology on sediment transport is actual based on the FLO-2D's one).

5) It is mentioned that the roughness coefficient has been assigned based on the land-use. Although this is the common practice, for the main stream, it is recommended to assign a different value (see recommendations by Chow, ).

6) Please consider fixing Figures 6, 10, 11, 12, since they seem to be stretched.

References

Dimitriadis, P., A. Tegos, A. Oikonomou, V. Pagana, A. Koukouvinos, N. Mamassis, D. Koutsoyiannis, and A. Efstratiadis, Comparative evaluation of 1D and quasi-2D hydraulic models based on benchmark and real-world applications for uncertainty assessment in flood mapping, Journal of Hydrology, 534, 478–492, doi:10.1016/j.jhydrol.2016.01.020, 2016.

Horritt, M.S., and P.D. Bates, Evaluation of 1D and 2D numerical models for predicting river flood inundation, J. Hydrol., 268, 87–99, doi:10.1016/S0022-1694(02)00121-X, 2002.

Merwade, V., F. Olivera, M. Arabi, and S. Edleman, Uncertainty in flood inundation mapping: current issues and future directions, J. Hydrol. Eng. 13(7), 608–620, 2008.

Mishra, A., S. Mukherjee, B. Merz, V.P. Singh, D.B. Wright, G. Villarini, S. Paul, D.N. Kumar, C.P. Khedun, D. Niyogi, G. Schumann, and J.R. Stedinger, An overview of flood concepts, challenges, and future directions, Journal of Hydrologic Engineering, 27(6), p.pp. 03122001, https://doi.org/10.1061/(ASCE)HE.1943- 5584.0002164, 2022.

Neal, J.C., I. Villanueva, N. Wright, T. Willis, T. Fewtrell, and P.D. Bates, How much physical complexity is needed to model flood inundation? Hydrol. Process. 26, 2264–2282, 2012.

Author Response

Literature has been read and have included some of it in the introduction- The comparison between models has been a significant issue of debate in the scientific fraternity [16-17]. The resulted differences are attributed mainly to the quality of topographic and input data [18] and less to the complexity of the phenomenon itself [19]. Several studies have compared the performance of 1D and 2D hydraulic models for river flood simulations [16,20] and have concluded that all models have proven sufficiently accurate, but have different responses when changing the friction parameters. Furthermore, they emphasize the fact that, no matter the quality of the input data, provided the user does not properly ﬁt the data into the appropriate geometrical description of the model, the ﬁnal results of the simulation will be considerably of lower accuracy [16].

The section of methods clearly narrates all this- explained in 2.2

Have also made some additions in the introduction - the last paragraph to cater for this as well

These are provided in 3.2. However the identification of possible sources of uncertainty and uncertainty analysis; were beyond the scope of this study

FLO- 2D Has been included in the most widely used models

Recommendations have been read but maintained earlier consideration. The coefficient values assigned are sufficient

6) Please consider fixing Figures 6, 10, 11, 12, since they seem to be stretched.

Figures 6, 10, 11, 12 have been accordingly fixed

Round 2

Reviewer 1 Report

All observations were clarified / corrected

Author Response

All observations were clarified / corrected

Thanks for the feedback

Reviewer 4 Report

The authors have replied to all comments; please consider addressing the 3 comments below following their replies from the first review:

1) In their reply, the authors mention that "Several studies have compared the performance of 1D and 2D hydraulic models for river flood simulations [16,20] and have concluded that all models have proven sufficiently accurate, but have different responses when changing the friction parameters.". However, this is not entirely accurate. Please see (for example) the review studies of Merwade et al. (2008), Dimitriadis et al. (2016), and Mishra et al. (2022), as well as many others in the literature that have already covered the issue of the important parameters in hydrological and hydraulic flood modelling, and clearly mention that friction is only one of the parameters with high sensitivity.

2) The authors replied that "The section of methods clearly narrates all this- explained in 2.2"; however, in 2.2 and 2.3 sections, I could not find the replies to:

a) What values of mean daily temperature, solar radiation, and wind speed, were considered for the estimation of the evapotranspiration.

b) Please show the land-use map.

c) What type of boundary conditions did the authors use in the steady-flow 1D HEC-RAS model?

d) Were there any of the input parameters linked to high variability or uncertainty (causing one of the models to under/overestimate the low or the peak flows and why).

e) The sentence "Steady flow analysis was used instead of unsteady flow analysis because in the second case, the HECRAS software needs a hydrograph, which we could not obtain from the local authorities. Thereby, to overcome this limitation, we used the flow rate for the gauging station." is confuding to me. Did the authors use the monthly flow from an upstream gauging station to calibrate the model based on the flow from a downstream gauging station? Please give more details on how exactly the calibration-validation-confirmation procedure was made.

f) It is not clear what the answers are to the questions raised by the authors, i.e., (1) how will the coupled hydrology-inundation model perform in order to attain proper flood assessment in a data-scarce area? and (2) can the developed model-based flood assessment fill the knowledge and technical gap in the development of flood protection measures and flood risk management? Please direct me to where these answers are and please further explain how the authors' research answers these questions for the area of their application.

3) What do the authors mean by their reply that "However the identification of possible sources of uncertainty and uncertainty analysis; were beyond the scope of this study.". The present study includes (even in the title) a risk assessment, which requires identifying (or at least discussing) sources of uncertainty in the input hydraulic (not just hydrological that are indeed mentioned in 3.2) parameters (e.g., besides the works mentioned by the authors, please see discussions and reviews in Merwade et al., 2008; Dimitriadis et al., 2016; and Mishra et al., 2022). For example, the size of the grid (it is mentioned 100 m by the authors), and the roughness coefficients at the floodplain and river-banks (it is mentioned that are between 0.01 and 0.4; where are these values originate from?) are only some of the sources of uncertainty that can highly affect the output (i.e., depth and velocity). These hydraulic input parameters are related to sections 3.3 and 3.4, where it is my understanding that no sensitivity analysis was made for the hydraulic modelling (but rather just for the hydrological model whose results are shown in section 3.2).

4) Thank you for addressing this.

5) I do not understand the authors' reply that "The coefficient values assigned are sufficient.". Are sufficient based on what? Please use (or at least discuss) how in the river and river-banks different roughness coefficients should be applied (see recommendation on the Manning coefficient in the famous work by Chow, 1959).

6) Please change Figure 6, since it still seems stretched.

Author Response

Several studies have compared the performance of 1D and 2D hydraulic models for river flood simulations [16,20] and have concluded that all models have proven sufficiently accurate, but they still have discovered that flood inundation modelling involves several sources of uncertainty such as: 1) input data (boundary and initial condition data, digital elevation models and channel bathymetry, hydraulic structures, roughness parameterization), 2) model structure (1D, 2D, quasi 2D, 1D/2D), 3) internal model parameters..

2) The authors replied that "The section of methods clearly narrates all this- explained in 2.2"; however, in 2.2 and 2.3 sections, I could not find the replies to:

a) What values of mean daily temperature, solar radiation, and wind speed, were considered for the estimation of the evapotranspiration.

Relative humidity, wind speed, solar radiation and the minimum and maximum air temperatures were obtained from the Climate Forecast System Reanalysis (CFSR), which was designed based on the forecast system of the National Centers for Atmospheric Prediction (NCEP) from 1981 to 2013 https://globalweather.tamu.edu. These are the input data and SWAT uses these to simulate evapotranspiration

b) Please show the land-use map.

land use maps have been inserted

c) What type of boundary conditions did the authors use in the steady-flow 1D HEC-RAS model?

River discharge data was used as the upstream boundary condition while normal depth was used as the downstream boundary condition.

d) Were there any of the input parameters linked to high variability or uncertainty (causing one of the models to under/overestimate the low or the peak flows and why).

The discharge and rainfall data are the main sources of uncertainty. The discharge data had many gaps and the rainfall data from the available weather station could not be used because it also had many gaps. The rain gauge network of the area is very sparse and as such the precipitation data was downloaded from CHIRPS for the 1981-2013 period.

For the Hydrodynamic model the sources of uncertainity was DEM because of the resolution

e) The sentence "Steady flow analysis was used instead of unsteady flow analysis because in the second case, the HECRAS software needs a hydrograph, which we could not obtain from the local authorities. Thereby, to overcome this limitation, we used the flow rate for the gauging station." is confusing to me. Did the authors use the monthly flow from an upstream gauging station to calibrate the model based on the flow from a downstream gauging station? Please give more details on how exactly the calibration-validation-confirmation procedure was made.

Calibration was done using monthly flow from an upstream gauging station. However due to lack of flood records validation was not possible.

f) It is not clear what the answers are to the questions raised by the authors, i.e., (1) how will the coupled hydrology-inundation model perform in order to attain proper flood assessment in a data-scarce area? and (2) can the developed model-based flood assessment fill the knowledge and technical gap in the development of flood protection measures and flood risk management? Please direct me to where these answers are and please further explain how the authors' research answers these questions for the area of their application.

Has been replaced with- In this study, we aim to address one scientific question: (1) how suitable is the coupled hydrology-inundation model for producing probability maps of flood plain areas for mapping vulnerability and risk in a data scarce area?

The discussion has a paragraph- It should be kept in mind that uncertainties exist in every stage of flood hazard mapping, from the beginning of the process (data collection, model selection, parameter selection, input data, model calibration, operation and handling of the models) until the outcome is obtained [39]. The main limitations of this study were data quality and availability (e.g., missing rainfall and hydrologic data; unevenly distributed discharge and water level gauges with varying time series length and missing data; little field survey cross-section data and lack of hydraulic structure data along the river, such as bridges, weirs, etc.) contributing to uncertainties and inaccuracy in the results. The accuracy of the flood maps could be improved through the identification of possible sources of uncertainty and uncertainty analysis; the sources of uncertainty include the DEM resolution. The current 30m is not sufficient and could lead to errors. We, therefore, recommended this for further research, as well as the integration of better quality data into the models if they become available.

4) Thank you for addressing this.

Thereby, the Manning roughness coefficient (n) was calculated based on land use/ cover classes in combination with typical roughness coefficient tables for each cross section, stream centerline, and river bank intersections with values (built-up area: n = 0.3; farmland: n = 0.025; bushland: n = 0.035; tropical high forest: n = 0.1; woodland: n = 0.06; wetland: n = 0.04 [31].

6) Please change Figure 6, since it still seems stretched.

The figure has been adjusted accordingly