the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Hydrological regime of Sahelian small water bodies from combined Sentinel-2 MSI and Sentinel-3 SRAL data
Mathilde Fleury
Laurent Kergoat
Manuela Grippa
Abstract. In the Sahelian semi-arid region, water resources, especially small water bodies such as ponds, small lakes and reservoirs in rural areas, are of vital importance. However, because of their high number and the scarce in situ monitoring networks, these resources and their spatio-temporal variability are poorly known at the regional scale. This study investigates the hydrological regime of 37 small water bodies, located in Mali, Niger and Burkina Faso, in Central Sahel. We propose a method based on remote sensing data only, which consists of combining water height data from Sentinel-3 SAR Radar Altimeter (SRAL) with water area data obtained with Sentinel-2 Multispectral Instrument (MSI) to create dense water height time series. Water height variations are then compared to evaporation estimated by the Penman--Monteith method using ERA5 reanalysis by the European Centre for Medium-Range Weather Forecasts (ECMWF) to infer water regimes during the dry season. Three main regimes stand out: a net water loss, mainly resulting from anthropogenic withdrawals, a net water supply occurring after the end of the rainy season through river network or water table exchange, and a balanced behaviour, where water losses during the dry season closely correspond to evaporation rates. Spatial patterns have been identified: in central Burkina Faso, most of the reservoirs show a net dry season water loss, which is explained by frequent irrigation, while reservoirs in northern Burkina Faso, generally show little water loss, indicating that water withdrawal is not significant in this area. Lakes located in the Inner Niger Delta in Mali and connected to the Niger river network generally show an important water supply, particularly at the beginning of the dry season. Lakes in Niger tend to show a weak signal toward water inflow that could be explained by exchange processes with the groundwater. These results show that satellite data are effective in estimating hydrological regimes as well as the anthropogenic impact on water resources, at the large scale, including resources found in small water bodies.
Mathilde de Fleury et al.
Status: final response (author comments only)
-
RC1: 'Comment on hess-2022-367', Anonymous Referee #1, 31 Jan 2023
The paper investigates the hydrological regime of 37 small water bodies, located in the Central Sahel region. This study is based on remote sensing data (Sentinel-2 MSI and Sentinel-3 SRAL) from 2016 to 2021, meteorological variables from ERA5 reanalysis, as well as some assumptions that help complete the dry season annual residual water balance for each lake. The residual water balance approach provides very valuable information on surface water resources at the regional scale. The problem that is discussed by the paper is rather interesting, particularly for an audience of developing countries in the West African Sahel region, where these sorts of studies are not really abundant. The monitoring and understanding of lake hydrological regimes are an important step toward better management of water resources. Therefore, the use of remote sensing for estimating the physical characteristics of reservoirs is potentially the most cost-effective management option in areas that suffer from serious security issues and high costs of in-situ data collection procedures, whose quality is not always reliable.
Overall, I think the paper is well written and of high interest to the readers of the Hydrology and Earth System Sciences (HESS). The methods and results presented are quite interesting and the efforts deployed for the analysis deserve full consideration. However, I regret that the authors did not validate some of their results, such as the relationships between water heights and water areas of reservoirs with ground truth monitoring data (field measurements) or initial A-H curves.
In addition, the following minor comments should also be addressed:
- Page 5, line 118: Equation (1)
There is a problem in equation (1) regarding the sign of evaporation in the residual water balance. Precipitation and evaporation fluxes are considered as an inflow and outflow respectively, therefore cannot have the same sign in equation (1). Please check this. If so, I hope the results are not affected.
- Page 5, line 123
The authors mentioned that the relationship between the water heights and water areas of a reservoir is also called “hypsometric curve”. I am not sure that this relationship is interpreted in the same way as the hypsometric curve, please check.
- Page 6, Table 1
In Table 1, it would be interesting to associate the uncertainties on the average values of the different variables
- Page 8, lines 153-156
Please, rewrite both sentences to clarify the final water height time series.
- Page 10, line 175
Please, can you explain the negative values of daily evaporation from the reservoirs?
- Page 10, lines 179-180
Please, check the definition of the residual water balance of a reservoir.
- Page 12, Figure 6
Please add a legend. For the X-axis labels, it would be good to write the months in letters followed by the year
- Page 12, Figure 7b
In Figure 7b, please check the date of the image. Is it 6 February 2018 or 21 January 2021?
- Page 13, Figure 8c
In Figure 8c, please check the date of the image. Is it 9 December 2019 or 4 November 2019?
- Page 13, Figure 9
Please add a legend. For the X-axis labels, it would be good to write the months in letters followed by the year
- Page 15, lines 224-225
This sentence needs to be clarified. Are these the watersheds of the lakes studied?
- Page 16, lines 239-240
Please, the authors should explain why the method developed is not sensitive to systematic errors in water body surface detection. Can the authors give some examples of errors? For example, the presence of waves in the reservoir is not discussed at all. Even if waves are not significant at the study area, it should be explicitly commented on given that the methodology is based significantly on the determination of water heights and water areas of reservoirs.
- Page 16, lines 240-241
The authors mentioned that “There is no need to derive an estimation of water area of the whole lake”. What would be the impact of underestimating the surface areas of water bodies on water heights? Could it be significant or not?
- Page 16, lines 260-261
Please, rewrite the sentence
There is a couple of references that are wrong!! (i.e. missing in the text of the manuscript, although listed in the list of references).
In my opinion, the main weakness of this paper is that the results obtained from the remote sensing data are not validated with field measurements even though the robustness of the tools and data used has been shown by some authors in other contexts.
Citation: https://doi.org/10.5194/hess-2022-367-RC1 -
RC2: 'Comment on hess-2022-367', Anonymous Referee #2, 14 Mar 2023
The paper investigates the hydrological regime of 37 small water bodies, located in Mali, Niger, and Burkina Faso, in Central Sahel, using remote sensing techniques of altimetry and surface water extent. It has been shown in previous studies that this is a valid method for assessing volume change in lakes. While the methodologies discussed are not novel, it is important to note that this study focuses on an understudied region of the Central Sahel. The authors point out that global studies do not quantify water fluxes over small water bodies in the Sahel, leading to several questions about small-scale hydrological regimes. The utilization of remote sensing techniques, particularly in understudied and under-gauged regions, has global significance. Thus this work is extremely important as water bodies such as these begin to have long-term monitoring.
An important finding of this study, focusing on smaller lakes over a longer time period than was previously conducted, is that the ability to study smaller lakes with a higher temporal resolution sensor enables the identification of more defined seasonal amplitudes and hydrologic variability.
I believe that this manuscript research is well designed, though not novel; the manuscript is well written, and the figures are polished, though the captions can benefit from more detail. I suggest some of the citations be double-checked, and there are some minor typographical errors. There is an uncharacteristically long paragraph at the end of the manuscript.
I give this manuscript an overall very positive review, and I hope some of the critiques and suggestions that I’ve made can be incorporated into the final revision. I'd like to thank the authors for an interesting and informative manuscript that will surely inspire more research.
Introduction
- Line 28 The authors write an interesting comment about the lack of information about temporary water bodies, but there is no defined (or written) methodology for segmenting temporary water bodies
- Line 39: Previous studies cite an RMSE of 0.67 m, which is quite large for water level accuracy, as systems are expected to have much higher accuracies, closer to 0.10 m. This study does not validate the water level measurements. The RMSE that is reported (Figure 3) is the RMSE of the model fit to the data (assuming the altimetry data is correct). This potential 0.67m difference is important when comparing results with other studies.
- Line 42: It is not recommended to cite Cooley et al 2021 for ICESat-2 or G-REALM as this group is not responsible for producing these data.
- Citations for G-REALM are suggested to include Birkett et al 2010 and Birkett et al 2017, and the USDA site directly [https://ipad.fas.usda.gov/cropexplorer/global_reservoir/Default.aspx].
- The sentence about laser altimetry is suggested to be re-written as ‘Studies have demonstrated using laser altimetry from ICESat-2 to measure water levels (Cooley et al 2021, (and others)).’ This suggestion is to clarify the relevance of that study to the present study, and not to confuse the production of the ICESat-2 data with the utilization of the data for the water level mapping study.
Section 3.2 Lake Water Height Estimation
- Please specify the frequency and sensor (Sentinel-3 Ku/C-band) used in AlTiS, to provide a reference for the backscatter coefficient.
Section 3.3 Surface Water Area Estimation
- It makes sense that the water index thresholds might vary by water body. Additionally, it would be a useful contribution to understanding the hydrological regimes if the mNDWI thresholds could be used to explain the variability between groupings of similarly reflecting lakes. (For example, grouping lakes where mNDWI==0, mNDWI> 0.1, mNDWI< -0.1)
Figure 2
- Please provide a description of the symbols in the figure caption.
Section 3.5 Evaporation estimation
- The Penman-Monteith equation does not require meteorological or DEM data from any specific source, such as SRTM or ERA.
- It is suggested to edit the sentence to say that the “Penman-Monteith equations…require the following meteorological and elevation data variables such as [var0, var1, var2, var3…]. Here, the meteorological variables are identified from ERA5, and the DEM is from SRTM.”
- The last two sentences mention the rainy season timing; it is not clear how that relates to the discussion of the input variables since the seasonal timing was not mentioned previously in the context of the P-M equations.
Section 4
- That 69% of the lakes turned out to be temporary lakes is interesting. The authors mentioned the importance of studying temporary lakes at the beginning of the article, but there was no mention in the lake identification section or in the methods that described how these temporary lakes would be identified or assessed. The majority of the lakes are temporary, it would be useful to have more of a description of them, perhaps in Section 3.3., where the spectral properties of the lakes are being assessed.
Section 4.1 Five-year averaged residual water balance
- Is it possible to add the colored boundaries from figure 1 to figure 5? This would make it easier to reference rather than going back up to the previous figure.
Figure 5
- Please provide a description of the symbols in the figure caption.
Figure 6
- Please provide a legend in the figure as well as a description of the symbols in the figure caption.
Figure 7
- Please provide a description of the shaded region in the caption (assuming it is the identified open water extent).
Figure 8 and Figure 9, the same comments as Figures 7 and 6 (requesting more labels).
Section 5 Discussion
- Lines 225-228: “Lakes also differ in terms of their characteristics…This complexity results in uncertainties in water regime calculation (Eq. 1).” I suspect that this comment might better relate to the mNDWI calculation where the water bodies are identified or the albedo calculation. It is true that the water surface variability or uncertainty will lead to uncertainty in the water balance, but it is important to point out that the source of the uncertainty is the mNDWI water identification (Eq 2), not in Eq 1 itself.
- Line 240 “Since areas are used to interpolate water heights..there is no need to derive an estimation of water area for the whole lake.” Please elaborate on or clarify this sentence. If the H-A hypsometric curves are developed based on water body areas, how is it possible that the whole lake is not necessary for this assessment? Does this mean that half a lake can be used as long as it is always the same half of the lake (an underestimate of the total area) that is developed in the H-A curve?
- Line 253. I would suggest saying that the retracking algorithm was designed for an ice surface, reducing its applicability to this task, as noted previously by Cretaux et al. 2018. (Not calling it a failure of the algorithm).
- Is page 17 (Lines 273-304) a single paragraph?
- It seems that this could be split into 2-3 paragraphs on water balance and ground water, and anthropogenic effects and the management of reservoirs
- Lines 259-261: Because the RMSE of the Cooley et al. 2021 study, and the RMSE of the altimeters used here are not included in the text, it is not possible to determine if the differences in the medians (2.06m this study; 1.60m Cooley et al. 2021), is significant. This idea is related to the earlier mentioned concept on line 39, where studies have asserted RMSEs of 0.69m. If that is the case then the differences in the medians between this study and Cooley et al 2021 may be within the margin of error.
Citation: https://doi.org/10.5194/hess-2022-367-RC2
Mathilde de Fleury et al.
Mathilde de Fleury et al.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 301 | 146 | 7 | 454 | 1 | 1 |
- HTML: 301
- PDF: 146
- XML: 7
- Total: 454
- BibTeX: 1
- EndNote: 1
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1