Review Reports - Anthropization and Climate Change: Impact on the Discharges of Forest Watersheds in Central Africa

Round 1

Reviewer 1 Report

Please, see the enclosed file.

Comments for author File: Comments.pdf

Author Response

Manuscript ID: water-908490

Type: Article

Title: Anthropisation and Climate Change: Impact on the Discharges of Forest Watersheds in Central Africa

Authors: Valentin Brice Ebodé, Gil Mahé, Jean Guy Dzana and Joseph Armathé Amougou

Reviewers of the article water-908490 made comments (major and minor) and suggestions to improve it significantly.

On this basis, several major corrections have been made to this new version of the manuscript.

The abstract has been reorganized, as recommended by reviewer 1. A sentence introducing the thematic has been added (Lines 13-15), as well as a short paragraph at the end of the conclusion which explains the possible future implications of the findings ( lines 34-35).

The references proposed by reviewer 1 (Yira et al., 2017; Namugize et al., 2018; Gorgoglione et al., 2020; Getahun et al., 2020) have been added to justify the statement "Most of the studies ... interferences "(Lines 44-45).

The aim of the work has been better explained in the introduction, as recommended by reviewer 1 (Lines 87-89).

The definition of the terms of the correlation coefficient calculation formula has been made, as recommended by reviewer 1 (Lines 240-241).

The word "interannual" has been replaced throughout the manuscript with "whole period" and "long-term", as recommended by reviewer 2. The words "variation" and "evolution" have also been replaced with "change ". The misuse of the word "trend" has been corrected. For example, the expression "identical trends were noted" has been replaced by "the same sign of anomalies was observed" (Line 465).

The results have been modified as recommended by reviewer 2, assuming that “variability” refers to the coefficient of variation (Cv) and that the percentages of variation shown are absolute changes in Cv.

Reviewer 2 wanted to know which definition of the runoff coefficient was used in this study because having observed a significant change in this variable from the 2000s. The definition of the correlation coefficient taken into account in this study is the same as his (ratio between river discharges and rainfall, when they are expressed in the same units). Their significant increase noted in this study in the case of Nyong is due to the fact that the rainfall has decreased considerably over the past two decades, compared to previous decades.

Reviewer 2 proposed to develop linear models based on interannual variations in rainfall and discharges, which could be used to predict (i) Spring discharges as a function of both Winter and Spring rainfall and (ii) Autumn discharges as a function of Summer and Autumn precipitation. We only developed a model based on the interannual variations of Summer rainfall (July and August) and the flows of the first month of Autumn (September), since it is only between these variables that the correlations obtained are good. For the rest, we have specified the values of the correlation coefficients obtained in the text, but we have not developed a model, taking into account that a model developed based on the data poorly correlated at the base cannot be used to make effective predictions. The complexity of the rainfall-runoff relationships in the region, due to evapotranspiration, changes in MOS, etc. could explain the poor linear relationship observed in some cases.

Several minor corrections were also made to the manuscript as suggested by reviewer 2.

The countries in which the watersheds are located have been specified (Lines 97-99).

The abbreviation ORSTOM has been spelt out (Lines 142-143).

The intervals of the rainfall correction coefficients have been specified (Lines 187-188).

Additional explanations relating to the graphics in Figures 3 and 4 have been added (Lines 335-337 and 340-342).

The use of bold font for some values in Table 5 has been explained (Line 480).

The period analyzed by Liénou et al (2008) has been specified (Line 393).

The section relating to the impact of changes in land-use patterns on runoff (3.3.2) has been slightly modified taking into account that no physical link was demonstrated between the two in the study. More careful wording has been done in some places in the text (Lines 595-635).

Corrections have been made to the language errors contained in the previous version of the manuscript as suggested by reviewer 2. These corrections are indicated in the current version, with the comment "Proposal for the second reviewer".

Reviewer 2 Report

GENERAL COMMENTS

The manuscripts analyses changes in precipitation and river discharge in two watersheds (Ntem and Nyong) in Cameroon in Central Africa from the 1950s to the 2010s, together with land use changes inferred from Landsat satellite images taken in 1973 and 2015. The major findings include (i) a significant decrease in annual discharges since the 1970s, despite no significant simultaneous decrease in annual rainfall, (ii) an apparent connection between the multidecadal variations of the wet seasons’ (spring and autumn) discharge with the preceding dry seasons’ (winter and summer) rainfall, and (iii) a strong increase in built impervious area at the expense of forest cover and water bodies, which might explain some of the discharge changes that are apparently not explained by variations in precipitation.

A strength of this study is the simultaneous consideration of precipitation and land use changes as drivers of river discharge change. However, an important limitation is that the analysis on the causes of the discharge changes is entirely qualitative, and the conclusions remain therefore speculative even if plausible. In addition, the statistical terminology that is used in describing the observed variations and changes is frequently unclear and unconventional, and the reading of the manuscript therefore requires some guesswork to understand what precisely is meant. The English language is reasonably good in the beginning and the end of the manuscript, but the middle parts including, in particular, the results section are plagued by a larger number of grammatical errors and cumbersome sentences.

For these reasons, major revisions seem necessary before eventual publication.

MAJOR COMMENTS

1. Section 3.3.1 makes several conclusions on how the rainfall changes in the preceding winter and summer seasons affect the discharge changes in spring and autumn. These claims would be more convincing if supported by regression models built on the interannual variations of precipitation and discharge. With several decades of data are available, it should be straightforward to build linear regression models that predict (i) spring discharge as a function of both the winter and spring rainfall and (ii) autumn discharge as a function of both the summer and autumn rainfall. With such models, the effects of the same-season and previous-season rainfall changes on the discharge changes could be objectively quantified. Furthermore, the residual time series from these regression models would have less interannual variability than the original discharge time series, making it easier to discern that part of the discharge change that cannot be explained by rainfall change.

2. Several statistical terms are used in an unclear, unusual or even erroneous way, which makes is substantially more difficult to follow the manuscript. In particular,

Table 1 and elsewhere: the word “interannual” is used to describe the long-term means of precipitation and discharge. Please change this to “whole period” or “long-term” and explain in the text that this covers the whole period of observations. “Interannual” is misleading because it usually refers to variation from year to year.
L297-298 and elsewhere: Please make it clear that “variability” refers to the coefficient of variation (CV) and the quoted per cent changes are absolute changes in CV. “Variability” might as well refer to the standard deviation, and the per cent change might be a relative rather than absolute difference between two periods.
Figure 3 and elsewhere. Please explain your definition of the “runoff coefficient”. I thought it was just the ratio between river discharge and rainfall (when expressed in the same units), but the very large changes in the runoff coefficient (in Fig. 3, more than a factor of two increase from the dry period to the wet period in Nyong) suggest that some other definition has been used.
Right column in Table 4 and elsewhere. Please replace “Variation” with “Change” or “Difference”. The word “variation” is misleading because it usually refers to shorter-term variability.
L430 (“identical trends were noted”) and elsewhere. This is misuse of word “trend” which generally refers to systematic changes with time. Please replace by “the same sign of anomalies was observed”
Right column in Table 7. Please replace “Evolution” with “Change”.

Please make these changes throughout the manuscript, not only in the specific places mentioned above.

MINOR COMMENTS / SUBSTANCE AND PRESENTATION, EXCLUDING PURELY LINGUISTIC COMMENTS

Please mention briefly the countries where the two watersheds are located.
What does “the latter” refer to? The Nyong and Ntem watersheds?
Which country? Cameroon?
Spell out ORSTOM.
L177-179. This short period of overlap for deriving the correction factors might be a major uncertainty in precipitation estimates after the year 2000. How large were the correction factors?
1 to t (not i-1 to t)?
when is sign(x) 0?
L228 and 231. Why “observations” and “estimates” (and which is which)? The definition of correlation is generic and does not depend on how the time series were obtained.
“neo-channels are created”: what exactly do you mean by this?
Is the “internnual average” just the mean value for the whole period for which observations are available (cf. Major comment 2)?
Its coefficient of variation decreases by 2% (cf. Major comment 2).
L319, caption of Figure 3. Please also explain the horizontal black and grey mean value lines and the baseline against which the wet phase and dry phase mean values were determined.
L333-334. These changes are negligibly small (which seems fortuitous, given the substantial interannual variability). The only thing that can be said is that the average maximums have remained essentially constant.
24 days earlier or 291-275 = 16 days earlier?
Table 4. The headings “maximum flows” and “minimum flows” have been switched.
L367-368. Please mention the period(s) analysed by Lienou et al., since differences in period may be important for the apparent differences in the results.
“never” = never earlier during the period covered by observations, i.e. since the 1950s?
Figure 4 suggests -9.7% rather than -8.2%
More precisely: … sources of long-term change and interdecadal variability
Table 5. Explain the use of the bold font. Does it indicate statistical significance of correlation?
Figure 5. What is the unit for precipitation anomalies? Are these normalized (anomaly in standard deviations) values?
years 1973 and 2015 for both Nyong and Ntem
Section 3.3.2. The reasoning in this section makes sense but the quantitative importance of land use changes remains highly speculative. As far as I see, the only way in which you could confirm your reasoning would be to run a hydrological model with and without land use changes. As far as this is not possible, more cautious wording would be appropriate in some places.
L648-651. Although this may well be correct, the wording would need some moderation because of the lack of quantitative analysis on the importance of land-use changes.

MINOR COMMENTS / LANGUAGE

It is hoped that the results of this study will
The annual precipitation oscillates
L103-104. On both sides of the main rivers, a topographic landscape in steps is observed which …
L104-106. In the middle part of each watershed, there is a major valley ( ), which …
during the 1990s and 2000s.
station rainfall
This choice
Indicators … were used
omit “different ranges”
L239-240. We can thus analyse not only the sign of change between the two periods but also the magnitude of this difference
L276-277. After validating the land cover maps produced, the statistical and spatial differences of each class between the two dates were studied ...
L295-296. a statistically significant decrease was noted in spring in both two watersheds
decline in them during the 2000s
L314-315. … discharges, although an increase in them is noted in the 2010s, after the decrease started in the 1970s.
their averages over the whole period
between the changes recorded
The minimums have occurred slightly later
L361-362. a slight decrease in the magnitude of these changes can be noted with increasing flow duration.
the decrease is greater for shorter-term minimums
L363-364. The coefficient of variation has increased slightly
L365-366. The minimums have appeared on the average 45 days earlier after the discontinuity
The change in annual rainfall (cf. Major comment 2)?
The 1950s and 1960s were wet on Nyong, then followed by …
L387-388. … opposite trends (Figure 4), with an increase in summer but a decrease in winter.
have led to the examination à motivate us to examine
This analysis
Opposite signs of NATL and the rainfall anomaly
years for which the same sign of anomalies was observed
the same sign of anomalies between winter rainfall indices
L437-438. Correlation between winter rainfall and SOI is good in both watersheds, but ..
years with the same sign of anomalies in winter rainfall and SOI
winter rainfall is correlated
indicates
L446-447. the negative correlation coefficient between
Frequency of years having the same/opposite signs of ocean indices and rainfall anomalies.
comparison of climatic indices and normalized rainfall anomalies
during the 1970s
rainfall does not decrease or decreases very slightly
This is all the more probable since
Omit ”noted”
rainfall does not decrease significantly, even though since 2005
Differences between these watersheds?
A few studies
linked to the size of the catchments, but also their large forest cover
until it reached 50% of its total area
L632-633. the proximity of the land-use changes to the outlets of the watersheds