Retrospective Predictions of Rice and Other Crop Production in Madagascar Using Soil Moisture and an NDVI-Based Calendar from 2010–2017
Round 1
Reviewer 1 Report
An excellent work overall, congratulations!
1/ It would be reassuring if further explantions were provided regarding how the following has been considered in this article: How rice production and distribution have been accounted for in this research, because there are the rice fields (tanimbary) i.e. the flooded ones, then the rainfed rice cultivation on hill (tanety). After rice has been harvested other crops are cultivated (such as potatoes). Also on the tanety the maize is in majority associated with rice cultivation. What about the growing of area under slahs-and-burn which increases over the years. And there is also the association of several corps including rice in an agroforestry system. All these can be well associated with soil moisture which is also characterize by the cultivation practice and the area where these crops are cultivated. These might be an explanation of the limits of the research that the reasons why there are insignificant relationships between soil moisture and rice production - while that is the crop that highly dependent on moisture/rain.
2) In order to support the data 42% of chronic malnutrition in Madagscar, it is better to use MICS 2018 as reference, because MICS cites what has been done with INSTAT which is more of the national official institution instead of WFP.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
-- The purpose of this study is to estimate rice, maize, cassava, and potato production in Madagascar from 2010 to 2017 using remote sensin. The manuscript's novelty derives from the research region (Madagascar) and the tools used (remote sensed soil moisture). The article, overall, requires significant revisions, including additional citations, descriptions of the research region and crop understanding, improved interpretations, and explanations of the results, as well as other adjustments recommended below.
--Specific feedback:
Please consider a more specific title. At the moment, the title does not adequately convey the content.
Overall, while the introduction discusses the primary rationale for this study (Malagasy drought stress), it is unclear on the methods and materials used. Please include information about prior attempts addressing this issue using MODIS, surveys, and near-surface moisture.While being an extremely concerning fact, nothing else in the manuscript is referring to the season 2020-2021. Consider giving context information about the time period covered by the supplied ground truth data (2010-2017).
Given that climate change is a keyword in this publication, please offer background information and reference on this as well.
L 25 – 28: This statement might be reinforced by broadening the scope to include all smallholders and include further reference.
L 31: stress may be a more apt description of the effect of weather on crop production.
L 31-33: Please offer further reference to support this statement and expand this thinking to incorporate additional information on crops and risk mitigation initiatives.
L 33: although the literature about Madagascar is sparse, numerous academics have published on this subject. Perhaps rephrasing this statement to go from the broad to the specific might help improve the text.
Please provide further citations for L36-39.
Figure 1: I'm assuming this is an issue caused by the Latex rendering process, so please ensure that figure 1 is repositioned to prevent interfering with the introduction. Include north and scale; moreover, consider modifying the NDVI color palette to red-green to emphasize areas with no or little vegetation (rocky central portion of the Country).
L40-44: Because these are the paper's objectives, they should be more explicit. Additionally, you might want to include this towards the end of the introduction.
L47 -60: suggest moving this section to the beginning of the introduction to offer further context, particularly on the usage of soil moisture and NDVI. Please include more citations to strengthen some of the paragraph's strong arguments.
Materials and methods:
Overall: Include a section explaining the region under study, the soils, the topography, and the climate's characteristics. All of these will aid in comprehending the dispersal of vegetation. While using MODIS for regional analysis is beneficial, it may obscure natural regions with a greater vegetation signal that are not always associated with food crops.
L65: briefly outline the significance and limits of NDVI (maybe in the introduction? ), as well as the main citations and further information about the use of MOD13.
MODIS: Why? Please include an explanation for your selection.
L67-69: What are the constraints associated with its use? What are the drawbacks of estimating crop production with surface moisture? Please provide further detail.
L74-76: Did the authors carry out this task? If this is the case, please clarify and include a source to support your technique. Is figure 1 displaying a spatial resolution of 0.25 degrees or 0.05 degrees? Additionally, this section of the paper should include figure 1.
L82-83: this might be included as part of the section on the area under study; moreover, consider adding a map with the districts. Is the population dispersed evenly across the country or is it more concentrated in some areas?
L101-103: this statement is not adding to the overall analysis. Although this is interesting and may be the explanation, given that the FAO statistics does not indicate this, I would recommend excluding the statement.
L106-110: How are these surveys performed (please provide further information)? During data collection, human bias can play a significant role. See: See: Lobell, D. B., Di Tommaso, S., Burke, M., & Kilic, T. (2021). Twice Is Nice: The Benefits of Two Ground Measures for Evaluating the Accuracy of Satellite-Based Sustainability Estimates. 13(16), 3160. Remote Sensing
Figure 2: it is not immediately evident what it depicts. Additionally, the legend bar is missing the final digit. Panels e–h includes exceedingly difficult-to-isolate data. What is a portion of harvested area?
L131: is this crop fraction?
L139: consider adding information as well as moisture during the critical period for each crop. This, in conjunction with water availability at the sowing date, is more relevant and has a bigger impact on yield components.
L180: why?
Where is this covariance given in L190-191?
L217-219: please provide a citation for this statement.
Results:
Overall, the results should be improved. Please include low correlation values as well; they are just as relevant as high correlation values. It is advised that more metrics associated with error should be included.
Additionally, a reminder of the sample size for each crop at each climatic zone would be beneficial, particularly to explain the observed variations. Certain crops will not thrive in certain areas because to their characteristics, and hence will be less likely to be planted in particular climatic zones. This fact may introduce bias into the analysis, and it should be addressed, similarly to how potato was discussed.
Table 2: many of the intervals are rather wide. Is this addressed in the manuscript? This might be a significant issue for certain crops and calendar/climate zones. Some include a considerable uncertainty. For example, C3-Wet-Maize, cassava, and others.
Please add a note on the uncertainty in L263-270.
While this is accurate for the national value, it does not reflect what is displayed for maize (in three of four areas), cassava, and potato.
L288-293: this section should be included in the discussion/conclusion.
Discussion-conclusion: Could be improved. Citations are a major requirement. Numerous arguments made in this discussion are well established in the literature and can be supported by increasing the number of citations in general and with an emphasis on agronomy.
L354-374: this belongs to the results.
Supplementary material:
Consider adopting the same scale for all productions and enlarging the graphs. This applies to all the figures in the document. For the untrained eye, this could be confusing.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
This manuscript shows a method for predicting the production of agricultural products in Madagascar from the amount of soil moisture, and I understand that it has great social significance. However, it is a slightly redundant manuscript including supplements, and it seems that there are too many explanations and data that are not necessary for a submitted paper. In addition, the explanation of the method progresses while showing the data of all crops, and the explanation and the result of the method are mixed in one chapter, so the content is too much and fragmented. As a result, it is difficult to capture logical continuity. Wouldn't it be more effective to sort out the information and carefully select and show only what is really necessary to give a logical explanation? I recommend that you make such corrections and resubmission.
P.3 L86-102 The reason why the authors focused only on the term of 2010-2017 was explained in detail including Fig S1-S3 in the supplementary but too long. What the author wants to express was that reliable data of crop productions were obtained after political crisis. The chapter can be shorten only with the reference[20]
P.4 Figure2 the numbers and characters on axis and legends were too small. No scale and northing arrow on the maps.
The production of maize was comparatively low. Better to use another legend. The color bar given in the manuscript could not indicate the spatial difference significantly.
P.4 L120-138 Explain kinds of crop distribution maps and selected the “crop-specific map” based on the high value of the correlation coefficient in Table 2.
My understanding from the tables, different from yours:
Regarding rice, the values are not so different from Table S2(best?) and S3.
On the other hand, the other 3 crops are tolerant to drought, as the author noticed. That’s why those shows low correlations except arid regions. In case of arid region, “Low SM” years are extremely so low that even Cassava was damaged.
P.6 L228 Equation (1) The author shows only the correlation coefficients but without the distributions between SM and P, so that I just guess from the surrounding data. Regarding the drought tolerance of the 3(or 4 including upland rice) crops, P might be assumed as constant (or with lower value of “a” in eq.1) when SM is larger than a certain threshold. When SM is smaller than the threshold, P might be expressed as eq.1. Showing the scatter plots might better understanding. The yield increment might have 2 phases.
P.7 Figure 3 C1 and C2 are not needed. I guess C2 was estimated from temperature and altitude. That would be standard but local farmers add their arrangement depending on the local characteristics. C3&C4 reflect that reality. If you omit C1 and C2, the volume of the tables and figures would be shrink totally. The comparison only between C3 and C4 indicates the impact of water stress near to harvesting season.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
The manuscript has improved significantly, but there are still a few elements to be addressed.
L25: What are the writers referring to when they speak of "agricultural shocks"? Please consider a more self-explanatory term.
L50: Please specify the nature of the resolution (temporal, spatial, etc.).
L117–133: Please reorganize this information. The remainder of the paragraph presents survey data, administrative division data, and average district size, all in one place. This has a detrimental effect on the readability flow.
L149: It is still unclear to me why this sentence should be retained. The explanation offered in the document, before and after, demonstrates the rationale behind it, which is backed up by facts. This is more than enough for the scope of this manuscript. Please remove this statement.
The discussion requires further work.
L413–435: this section of the discussion goes too deep into the results.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
While the frequency of droughts is increasing around the world due to climate change, Madagascar, which has an extreme climate distribution in a small island due to the influence of the monsoon and morphology of the island, is susceptible to droughts and food production is liable to fluctuate significantly. .. I got that the authors are acting for the prediction of the production of staple food.
While the frequency of droughts is increasing around the world due to climate change, Madagascar, which has an extreme climate distribution in a small island due to the influence of the monsoon, is susceptible to droughts and food production is liable to fluctuate significantly. .. On the other hand, it is important to predict the production of food, especially staple food, and it is fully understood that the authors are serious about it.
Given the simplicity of the factors that form the climate distribution, there is no dispute that soil moisture is a major factor in abundance variability in crop production. Perhaps it is a Madagascar-specific evaluation method rather than a global evaluation method. I think it deserves high praise as a highly accurate method that is very practical in consideration of regional characteristics.
The authors present a very large amount of data, and easy to see that the process of data processing is explicit, but it still feels encouraging to me. The conclusion in this paper is that soil moisture can express the total production of staple foods under conditions of relatively simple climatic distribution. Am I the only one who thinks that simply showing this simple solution will increase the value of the manuscript? I think the conclusions reached are very valuable. On the other hand, it may be the role of the Editor, not me, to decide whether or not to devote more than 30 pages including Supplementary. Accept as peer review.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
