Response of Grain Yield to Extreme Precipitation in Major Grain-Producing Areas of China Against the Background of Climate Change—A Case Study of Henan Province

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This manuscript investigates an important and timely issue: the impact of extreme precipitation on winter wheat yield in Henan Province, China. The authors employ a comprehensive suite of statistical methods, and the identification of multi-scale relationships offers a valuable contribution to the field. The study is generally well-conducted. A few clarifications and enhancements, detailed below, are recommended to further strengthen the manuscript before publication. 

 

General Comments

The manuscript investigates the impact of extreme precipitation on winter wheat yield between 2000 and 2023 in Henan Province, China, a topic highly relevant to the scope of Water. The authors are commended for their use of multiple statistical techniques, including Mann-Kendall trend tests and various wavelet analyses (WTC, XWT), to explore the temporal and spatial evolution of extreme precipitation indices and their multi-scale stress mechanisms on grain yield. The study identifies key characteristics of extreme precipitation, changes in yield stability, and proposes a "sub-scale governance" strategy. The following suggestions are offered to improve the manuscript's clarity and impact.

Points Requiring Minor Revision

Clarification of Winter Wheat Yield Data: In Lines 170-178 authors states that wheat data used is the "30-meter resolution planting distribution V2 data set of winter wheat in China from 2001 to 2024 (http://www.nesdc.org.cn)." It further describes this as a "spatial resolution winter wheat planting distribution map." A  planting distribution map indicates the spatial extent or location of wheat planting, not the yield (e.g., kg/ha or tons). The study, however, analyzes "winter wheat yield" throughout (e.g., Abstract, Fig 6 showing "total output of wheat," wavelet analyses correlating indices with "yield"). This is a fundamental methodological ambiguity so I suggest authors must explicitly detail how actual yield data (quantity per unit area or total production for administrative units) was obtained and integrated with this planting distribution map. Was county-level statistical yield data used and then masked or aggregated using this spatial distribution? Or is "planting distribution" being used as a proxy for yield, which would be highly problematic? The accuracy metrics (91.17%) refer to the identification of planting areas, not yield estimation.

Development and Justification of "Sub-scale Governance" Strategy: In Lines 638-647 where authors propose a "sub-scale governance" strategy with specific measures like "two-night double-optimal technology," subsurface drip irrigation, rainwater harvesting, and cascade storage. While these are known agricultural water management techniques, the link between the specific analytical findings of this paper (e.g., the 1–2-year lag effects, the dominance of R99p, specific mutation points) and the formulation of these particular strategies needs to be much more explicit and robustly argued in the Discussion and Conclusion.  How do the multi-scale stress mechanisms identified directly inform or necessitate these specific governance approaches over others? For example, how does the identified 1–2-year lag effect (Line 32) informs the "two-night double-optimal technology" (Line 639)? The term "two-night double-optimal technology" is not standard so I would like to suggest you to either do a citation or a more detailed explanation of what it entails within the text.

Interpretation of Specific Correlations: In Line 543 states: "It is worth noting that the number of 'continuous dry days' (CDD) in summer is positively correlated with yield." This is counter-intuitive at first glance, as prolonged dry days, especially in summer, are typically stress crops and reduce yields. The authors should in deed provide a strong, plausible agro-climatological explanation for this finding in the Discussion. Is it because longer dry spells reduce disease incidence often associated with wet summers, or prevent waterlogging damage that might be more severe than moderate water stress? Also in Line 550 : "For every 10% increase in summer extreme precipitation, the fluctuation rate of grain yield will increase by 1.2%". could be misinterpreted. It’s essential to distinguish between “variability” and “mean yield impact”. Does "fluctuation rate" mean variability increases, or does it imply a net increase in yield, which would contradict the general finding that summer extreme precipitation has a negative effect (Line 535)? So, it’ll be better if the authors could clarify if this refers to increased inter-annual variability rather than a direct positive impact on mean yield.

Data for Meteorological Stations: While NOAA global stations are mentioned (Line 161-162), for a regional study of Henan Province, it would be beneficial to specify the number of stations within or immediately surrounding Henan that were used in the analysis. This helps assess the spatial representativeness of the precipitation data.

 

Other Minor Suggestions

1. Clarity and Conciseness of Language: Some phrases used in the article are a bit clunky or overly technical for a broad audience, e.g., "high frequency oscillation-weak trend-strong spatial heterogeneity" (Abstract, Line 22; Conclusion, Line 599). While descriptive, consider rephrasing for smoother reading. "red production reduction plaque" (Line 622) is an unusual term. "Areas showing yield reduction" or similar would be clearer. Abstract (Line 26): "2010 was the key mutation node, which triggered the synergistic effect of 'drought aggravation-heavy rain increases'". This phrasing is a bit dense. Consider breaking it down.
2. Figure Clarity and Presentation:

1）Figures 2, 3, 4, 5, 9, 10, 11, 12: These figures combine spatial maps, time series plots, and UF/UB curves for Mann-Kendall tests. They are information-rich but also very dense. Consider if any could be split or simplified for better readability, or ensure high resolution for publication. And also try to make text more clearly visible in all the figures.

2）Figure legends are generally adequate, but ensure all components within complex figures are clearly explained.

3）Fig 6 (Winter wheat interannual): The y-axis label "The total output of wheat (ten thousand tons)" is clear. This again raises the question one  Issue #1: where does this "total output" data come from?

4）Fig 7: Legend "Wheat producing area" / "non-wheat producing area". This is consistent with a planting distribution map. The conflict with "yield" analysis remains.

3. Specific Textual Points:

1）Line 171: The winter wheat V2 dataset is cited as 2001 to 2024 [55], but the analysis period for wheat is stated as 2001-2023 (e.g., Line 339). Please ensure consistency or clarify if the 2024 data was used for a specific reason not apparent in the current text.

2）Line 252: "RX1DAY showed a significant upward trend in spring (slope = 0.6 mm / year, R2 = 0.631)". An R-squared of 0.631 for a trend like this is exceptionally high. Please double-check this value to ensure the accuracy and transparency.

3）Line 607: "the pattern of 'more in the south and less in the north' was highlighted spatially." This seems to refer to R99p. Is this pattern consistent across other indices or specific to certain types of extremes?

4）Data Availability Statement (Line 672): "The data presented in this study are available on request from the corresponding author due to (specify the reason for the restriction)." The reason for the restriction is missing and needs to be specified.

4. Discussion Depth:

1）If possible, discussion on the physical mechanisms behind the identified lag times (e.g., 1-2 years for RX1DAY affecting yield) could be deepened. Are these related to soil moisture memory, impacts on sowing conditions for the next season, farmer adaptation strategies, or multi-year effects on soil health?

2）The link between ENSO and Henan's precipitation (Lines 463-466) is mentioned. Could this be briefly elaborated in the context of the study's specific findings on mutation points or dominant cycles?

Author Response

All responses were attached in the Response to Reviewer 1

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Dear authors, I found your ambition to link multi-scale extremes in precipitation to winter wheat performance across Henan both timely and valuable. However, several methodological loose ends kept me hesitating as I read.

Comment 1: Regarding the data, you state that daily precipitation comes from the NOAA global station archive for 2000-2023 and extol its 70-year depth and multi-source verification, but you never enumerate how many stations actually fall inside Henan, how evenly they are distributed, whether any were excluded for missing values, or how you reconciled overlapping gauges with the gridded products sometimes bundled under the same NOAA umbrella. Without that basic station meta-information—and without a short description of the quality-control flags you retained or rejected—it is hard to judge the representativeness of the extremes you later derive from the data.

Comment 2: Similarly, the wheat panel is introduced as county-level yield data, yet the only explicit source I see is the 30 m V2 planting-distribution map. Does this map serve merely as a mask while the actual yield numbers come from statistical yearbooks, or are you inferring yields using remote-sensing proxies? If the latter, what calibration against ground harvests underpins the accuracy, and how is spatial downscaling handled when you later aggregate to the CV maps? In my opinion, you should clarify the yield provenance, which is critical, because otherwise the wavelet work risks chasing noise introduced by mixed data-generation pathways rather than genuine agronomic signals.

Comment 3: Furthermore, the Mann-Kendall tests are appropriate for distribution-free trend detection, but applying them to eleven indices in four seasons multiplies the family wise error rate. You cite field-significant work in other contexts, yet I do not see a correction (e.g., false-discovery rate or the regional MK variant) implemented here. Could some of the significant seasonal breaks simply reflect chance under multiple tests? Your wavelet coherence and cross-wavelet diagrams are visually good, but the manuscript never explains the choice of mother wavelet, the padding scheme outside the cone of influence, or the Monte-Carlo surrogate settings that determine the 95 % coherence threshold . Because you later interpret lags of one to two years as agronomically actionable lead times, I need confidence that those high-energy regions exceed red-noise expectations rather than artifacts of shared low-frequency trends.

Comment 4: Your interpretation itself sometimes slips from correlation into causality. For example, you attributed yield dips after 2010 to the “synergistic effect of drought aggravation–heavy rain increases” without adjusting for concomitant trends in temperature, fertilizer inputs, or cultivar turnover—all known co-drivers of wheat yields in your study area. I suggest adding a panel regression that partials out those confounders (or at least a sensitivity analysis), which would support the claim that precipitation extremes are the dominant driver.

Comment 4: Similarly, your conclusion advances prescriptive strategies (subsurface drip irrigation, stepped rainwater harvesting, and “two-night double-optimal” sowing); however, the body of the paper does not test or cite empirical evidence that these measures specifically mitigate the wavelet-identified stress periods. Please bridge that gap, as it would make the recommendations feel earned rather than aspirational.

Comment 5: Structurally, the introduction is thorough and well-referenced, the methods section covers each statistical tool, the results are extensive, and the discussion broadens to the implications before a concise conclusion. However, a few seams were observed. Your Section 3 sometimes slides into policy recommendations that belong later, while the discussion repeats numeric results already given in Section 3 instead of focusing on interpretation. I also missed an explicit “Limitations” section; many are implied, yet stating them up front would demonstrate reflexivity.

Overall, your core idea is strong, and the blend of trend tests with time-frequency analysis will make a real contribution if the empirical scaffolding is tightened. I encourage you to (i) document station counts, gaps, and any homogenization; (ii) spell out the yield data pipeline and validate it against the ground truth; (iii) apply multiple-testing controls to MK outputs; (iv) detail wavelet parameters and significance testing; (v) explore or at least discuss non-precipitation covariates; and (vi) relocate policy prescriptions to a clearly delineated section grounded in your findings. Addressing these points will not only bolster methodological rigor but also provide readers with greater confidence that the climatic signals revealed are robust and that the adaptation strategies proposed are evidence-based.

 

Author Response

All responses were attached in the Response to Reviewer 2

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

I appreciate the author's effort for revising the manuscript. Authors have properly addressed all my raised concern on original manuscript.

Reviewer 2 Report

Comments and Suggestions for Authors

thank you for considering all my comments. Best wishes