the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 12 Mar 2025
Assessment of source regions of the Zambezi River: implications for regional water security
Abstract. The Wilderness Project, in collaboration with the National Geographic Okavango Wilderness Project and the Wild Bird Trust, conducted the first known scientific river expeditions along the entire length of the Zambezi River—from its traditional source in Zambia and its most distal source in the Angolan Highlands Water Tower, all the way to the Indian Ocean. By combining ground measurement and high-resolution earth observation data, this study describes the overlooked contribution of Angolan Highlands Water Tower to the Zambezi River. Our findings show: 1. The Zambezi River length, measured from the Lungwebungu River, is 342 km longer (total length of 3,421 km to the Indian Ocean) than the traditional source, 2. We estimate that the Angolan portion of the Upper Zambezi and Lungwebungu sub-basins contributes 72.79 % of the flow measured upstream of the Barotse Floodplain, 3. The Lungwebungu and Upper Zambezi sub-basins reduce elevated conductivity, salinity, and TDS, likely introduced by mining activities in the Kabompo sub-basin, and 4. The Upper Zambezi sub-basin has the highest proportion of wetland coverage, with 94.61 % (19,184 km2) in Angola (specifically the Kameia and Luena wetlands), promoting river health. This study highlights the recognition Angola deserves for its critical role in hosting the source and primary aquifer of the Zambezi River. Additionally, it underscores the need for improved monitoring and analysis of hydrological flows of the Zambezi Basin, particularly its Angolan tributaries. The Zambezi River Basin faces ongoing challenges from climate change, development, and human water use. Collaborative efforts among basin countries are crucial to balance development with environmental needs, safeguarding ecosystem services and the natural dynamics that sustain the Upper Zambezi’s ecological integrity.
- Preprint
(3272 KB) - Metadata XML
-
Supplement
(2105 KB) - BibTeX
- EndNote
Status: open (until 07 May 2025)
-
RC1: 'Comment on egusphere-2025-837', Anonymous Referee #1, 18 Apr 2025
reply
Lourenco et al. through this contribution wish to clarify the hydrology, ecology and climatology of the Zambezi, as one of the most important basins in Southern Africa. Understanding these processes including their interactions is crucial, especially now under the current rapidly developing conditions in the basin. There is a large interest in mining, further use of water for food, and hydropower, all likely impacting in different ways on the Zambezi’s water resources. As seen from last year’s drought, exclaimed (in Zambia) to be the largest in 20 years time, climate variability will pose a large challenge as well. In order to protect, sustain and distribute water resources in an equitable manner, any developments should be underpinned by scientific data and analysis, that predicts the likely impact of the development in conjunction with climate change and climate variability.
Now did Lourenco et al. succeed in making a contribution to this end? I have quite struggled to get through the paper and to find the most prominent merits. The paper’s English is likely better than my own, but for a scientific contribution, there is too much data, there is also too much data that is not new and also not analyzed in a novel way, and many data also do not lead to new evidence or quantifications of hydrological processes or other key hydrological knowledge, which would make it a logical contribution for this journal. The paper reads as a report rather than a scientific article.
I start here with commenting on the 5 key findings from the discussion:
l. 613-614 “...reveal five significant contributions from rivers originating within Angola that have been overlooked”. This already sounds as if nobody ever thought of the fact that a significant portion of water from the Upper Zambezi comes from Angola. The well known and widely used HydroSHEDS dataset already shows that a very large portion of the upstream area comes from this area. This is quite trivial and none of the data collected and shown gives very new insights. On the 5 key points below that:
1. The length of the Lungwebungu tributary: the Lungwebungu may be longer than what is currently considered as main stem, but this length has long been known from terrain analyses. I do not see any new insights here. There may be other reasons (as the authors show in their own literature review) to call the origin of a certain stem the “source”, from a hydrological or water resources point of view, this is not relevant. It is only a narrative.
2. The quantification of the Angolan contribution. Probably it is the largest, but as said, is that really a novel finding? Also, the empirical gauge records used to support this, cannot provide any evidence whatsoever, as observations were only snapshots, and even not taken at the same moment. I understand that taking measurements at the same moment is not possible, (and appreciate how difficult it is to collect these) but this “new insight” is thus not supported by data.
3. Provided only data taken at the same moment (2023/05/21 for the Kabompo), and the fact that a 15 times higher concentration was found in the Kabompo, this finding is to my mind supported sufficiently by data. The Kabompo lies in the middle of the Copperbelt region, and is subject to a lot of mining activities, which may be the cause of this high salinity and other contaminants. The Lungwebungu concentration however, was taken during the rainy season, and this measurement is therefore difficult to compare. During this season, flows are likely (much) higher than the dry season, hence this can dilute contamination sources, rendering any conclusions on this matter false.
4. The proportion of wetlands. The authors only reduced the wetland map over subbasins, this is hardly a new finding. It is only a summary of already existing data sources.
5. The highlight of significance: this is not really a finding, merely a confirmation what was already known.
So all in all, despite the length of the paper, I find it not strong and not novel. To my mind, the strongest contribution lies in the water quality sampling. This was done in a season with relatively low flows, and with the order of magnitude difference in concentration measured, I think the conclusion that the Kabompo is negatively impacting the water quality of the Zambezi is robust. If this paper is to be accepted, I would recommend focusing on this finding, only provide detailed description of data and analysis that supports findings (you may mention which data you considered and then quickly conclude why they were not suitable), and deepening the water quality aspect out with further measurements that can also give insights in the variability of the quality in time. This may the lead to proper quantification and possibly scientific support for policy on further developments or conservation in the Kabompo basin, and/or development and conservation of the Angolan part as a contributor to improved water quality. The collaboration with the Zambia Water Resources Management Authority is crucial for this.
I also have provided more in-depth comments below.
l. 43. The river first traverses Southward over a very significant length. This is important to note as over that length also the ecology/climatology changes from equatorial forests with tropical climate to semi-arid Savannah.
Table 2. The line on empirical discharge suggests that in-situ ADCP measurements would result in “mean discharge”, which suggests mean in time. But that is not true. It provides only instantaneous discharge at the moment of the measurement. Such a measurement can never be used to conclude on accumulated contributions of a river tributary. See my main comments.
l. 290 onwards. How did you combine sampled ADCP observations with gauge records and the modeled records? What conclusions can you draw from that? This is quite important and a strong weakness in the paper. For instance, if you would take a end-of-dry-season low flow observation in the Zambezi main stem just below the Barotse floodplains, that will be relatively high (after dividing over the upstream area) compared to the low flow observations in faster responding systems such as the Luangwa. However, this does not say anything about the longer term contributions. Also, assuming you can use one single observation to scale or bias correct modeled flows is not a viable approach as hydrological models, that are not fully calibrated, will have seasonally varying biases. This comment also leads to my main comments.
l. 303. Gaps are filled with EO datasets, but from what is written underneath, it is not clear how you would end up with reasonably bias-free river discharge numbers. It only describes typically available terrain-derived variables.
L. 315 onwards: Modelled runoff by Akpoti et al and Collins et al. Likely the very gauge records the authors use to compare it against, have also been used to perform monthly bias-correction of the data, and likely the sole reason to utilize these data for this study is to estimate river flows in ungauged areas, where (trivially) no records were available to perform bias correction. Given the highly variable landscapes throughout the basin areas (also described by the authors themselves) it is also not likely that the bias follows a similar pattern (e.g. the Zambian upper Zambezi is characterized by thick Kalahari sands, while the Angolan water tower does not have large alluvial aquifers). Furthermore, especially in semi-arid parts of the investigated basins, a small error in the modeled water balance will result in a very large error in the runoff and discharge, as discharge is a non-linearly responding residual of the hydrological processes.
l. 335 “Given the limited availability of comprehensive water chemistry data across the Zambezi Basin, we have incorporated a wetland and surface water inventory for each HydroSHEDS Level 04 sub-basin to complement existing data and serve as a supporting proxy indicator of overall river and ecosystem health.” What do you mean by a “wetland and surface water inventory”. How do you use this to characterize ecosystem health?
l. 400. How were the errors of the river length calculated?
l. 435. “flow accumulation pixels”, please convert this into a metric unit such as km2.
l. 441. “Sananga” should be “Senanga”.
l. 458 – 466. The precision of the flow estimates and percentages contributions is unnecessarily and unrealistically high. Furthermore, it is unclear what can be drawn from the relative contributions during just one snap shot.
l. 490-506. Also a lot of very precise numbers. It is also not clear to me why rough estimates of percentage-wise contributions would have to rely on single snapshot ADCP observations. A much more logical proxy would be the catchment area multiplied by long-term runoff estimates using the Budyko curve. One snapshot just does not make sense as no seasonal variability or seasonal accumulation can be derived at all.
Table 8. The description “Seasonality: Number of months water is present within a calendar year, average 1984 – 2021.” is not clear to me. The table shows percentages for each individual month, not yearly aggregates.
Fig 7. Is subplot b not simply a differently scaled version of subplot a? If not please explain what both maps exactly mean and why both are required for your analysis.
l. 626 – 634. I do not see the scientific importance of the “finding” that the Lungwebungu is longer than the length from Lukulu to the source. It is merely a geographical characteristic.
l. 643 – 644 “Although the Angolan portion covers only 14% more area, it contributes approximately 45.58% more flow at the Lukulu confluence than the Zambian portion.” How is this concluded? If that comes from the snapshot ADCP observations, then the conclusion is not valid, (see main comment) as you would require multi-year time series to properly assess this.
l. 659 – 666. Not really something that belongs in the discussion, more logical for the introduction.
l. 669 – 677, also not a discussion based on this work, merely some text about the basin from existing references. It does not belong in this section to my mind.
l. 679 – 688. Also this contains no discussion points based on this work.
l. 708 “Its size is the reason for over-estimations of modelled discharge”. Why would the size result in an overestimation? It is more likely poor representation of semi-arid hydrological processes in the used models or biased rainfall.
l. 731. “pivotal—yet previously overlooked—implications for transboundary river management”. This sounds very strong again, as if no-one ever thought about the fact that a large part of the Zambezi’s water stems from Angola.
Citation: https://doi.org/10.5194/egusphere-2025-837-RC1
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 144 | 37 | 4 | 185 | 4 | 5 | 4 |
- HTML: 144
- PDF: 37
- XML: 4
- Total: 185
- Supplement: 4
- BibTeX: 5
- EndNote: 4
Viewed (geographical distribution)
| Country | # | Views | % |
|---|---|---|---|
| United States of America | 1 | 66 | 32 |
| South Africa | 2 | 39 | 19 |
| Botswana | 3 | 10 | 4 |
| China | 4 | 8 | 3 |
| Germany | 5 | 6 | 2 |
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
- 66