Exploring Trends in Earth’s Precipitation Using Satellite-Gauge Estimates from NASA’s GPM-IMERG

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This study, based on NASA's IMERG V7 dataset, provides a comprehensive analysis of global precipitation trends over the 27-year period from 1998 to 2024. Using non-parametric statistical methods (Mann-Kendall test and Sen’s slope), it detects and quantifies trends at annual, seasonal, and monthly scales. The research identifies key patterns, including wetting trends in high-latitude regions (the Arctic and the Southern Ocean) and drying trends in tropical areas (such as the eastern Pacific, northern Amazon, and central-southern Africa), linking these patterns to broader climate drivers such as polar amplification and ocean-atmosphere interactions.

The topic is timely and relevant to climate change research, particularly for understanding hydroclimatic changes. The use of satellite-gauge merged data helps address observational gaps in remote regions, and the pixel-based analytical approach provides valuable spatial insights. However, the manuscript has several limitations, including insufficient discussion of data uncertainties (e.g., IMERG biases in high-latitude regions), potential overinterpretation of trends from a relatively short record, and a need for methodological clarification. In addition, there are inconsistencies in units, figure labels, and citation formatting.

(1) IMERG V7 was used, but high-latitude biases (e.g., snowfall underestimation, sparse gauge calibration) are only briefly mentioned in the conclusions (lines 304-308). This needs to be expanded in the methods and discussion sections, especially since it may affect the wetting trends in the Arctic/Southern Ocean. Provide a deeper reference to recent validation studies (e.g., [28] Milani & Kidd, 2023).

(2) The record length (27 years) is considered a limitation (line 304), but it should discuss how this affects trend detection, particularly for low-frequency variability (e.g., through power analysis or comparison with longer reanalysis datasets like ERA5).

(3) No mention of potential artifacts from the TRMM to GPM transition (e.g., sensor differences before/after 2014). Clarify if any homogenization processing was applied.

(4) The Mann-Kendall test assumes data independence, but precipitation time series often exhibit autocorrelation. Did the authors consider this (e.g., using a modified MK with trend-free pre-whitening)? If not, justify or apply it.

(5) Sen's slope units are reported inconsistently (e.g., mm/day/yr in the text, sometimes mm/yr in figures). Standardize to mm/day/yr for clarity, as daily rates are more suitable for precipitation.

(6) The text mentions R for processing, but specific packages (e.g., raster, Kendall) should be specified to ensure reproducibility. Additionally, how were missing values or land/ocean masks handled?

(7) The statistical significance in the article is set at p ≤ 0.05, but it is recommended to consider field significance testing (e.g., false discovery rate) to address the multiple comparisons issue across pixels, reducing Type I errors in spatial data.

(8) The scatterplots in the article (e.g., Figures 3, 5, 7, 9, 11) show slopes for "regions with significant positive/negative tau values," but these regions need to be defined more precisely (e.g., bounding boxes or masks).

(9) The wetting/drying dual pattern is a key conclusion, but it needs quantitative comparison with other global studies (e.g., [13] Zhu et al., 2024; [16] Nguyen et al., 2018)—are the magnitudes consistent?

(10) There are some minor spelling/grammar errors in the article: e.g., “statically” → “statistically” (line 156); “As in with” → “As with” (line 199); “declining JJA precipitation (Figure 8 a)” → “Figure 8a”.

(11) Some 2025 citations in the text (e.g., [23,30,25]) are included; please verify publication status. Formats are consistent (e.g., some DOIs are missing).

Author Response

Please check the uploaded document. 

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Summary

This manuscript presents a comprehensive analysis of global precipitation trends from 1998 to 2024 using NASA’s GPM-IMERG dataset. It applies statistical methods, like Mann–Kendall trend detection and Sen’s slope estimation, at pixel-scale resolution to investigate the precipitation variability across multiple regions and seasons. The results reveal consistent patterns of high-latitude wetting (Arctic and Southern Oceans) and tropical drying (East Pacific, central–southern Africa, northern South America), with Northern China showing persistent increases. The manuscript is generally well-structured, well-written and scientifically sound. Some typos were found so I would suggest a final careful proofreading before proceeding. My comments are mostly focused on minor improvements related to figure presentation, readability, and the summary of results.

Abstract

The abstract is well written, clearly presenting the aim of the study, the dataset used, the applied methodology and the key findings. The inclusion of quantitative results adds strength and credibility. Well done!

Minor comment: Suggest using the same number of decimals (two points is adequate) for the quantitative results (e.g. +0.0484 mm/day/yr -> +0.05 mm/day/yr)

Introduction

The introduction section is well structured and comprehensive. It states the importance of global precipitation monitoring, the satellite missions of TRMM and GPM combined via the IMERG, sufficient relevant literature as well as the objective of the present study. Some minor comments:

1. Reduce repetition of IMERG capabilities and performance. Keep one concise paragraph on data quality and validation. For example, the 10% bias is mentioned in lines 47-48 and 57-58.
2. The last paragraph stating the objective of the work should highlight more of what is that is missing from past studies. The authors could end the literature review by stating what has not yet been done (e.g., consistent global pixel-scale trend analysis in multiple timescales) and that is why this study is conducted. This way the scientific value of this work is highlighted.

Data and Methods

The data and methods are presented clearly and the text flow is easy for the reader to follow. A description of the dataset is given, along with the methodological steps and statistical methods that are applied (Mann-Kendal, Sen’s slope). Well done!

My comment is about Figure 1. The Figure is mentioned but not discussed at all. I would suggest either moving the Figure in the Results and Discussion section and commenting there, or leaving it in the Data and Methods section, but at the beginning of the Results section add a few sentences describing the results of this plot about the spatial variability of precipitation. For example:

“The annual mean rainfall estimates (1998–2024) are seen in Figure 1. The highest values occur over the tropical belt, reaching up to 30 mm/24-hr, whereas northern Africa, inland Australia, and parts of the Middle East exhibit very low rainfall (<1 mm/24-hr).”

Results and Discussion

The results are well presented and are scientifically robust. My only comments are regarding some minor adjustments to the figures:

1. Figures 3,5,7,9,11: I recommend increasing the text in the Figure by a lot. Right now, is not easy to read. You have a plot with 4-6 different trends and not 4-6 different plots in one Figure which is good and gives you the space for bigger fonts. So, I would suggest increasing the size of the axis titles, tick labels, legend etc
2. Figures 4,6,8,10: Since (a), (b), (c) and (d) exist in the Figures, the same reference format should be used in the Figures’ captions (and not top-right, bottom-right, etc).
3. Since the figures are repeating for different timescales, I would recommend using a Figure caption like “Same as in Figure x, but for …”. This way you avoid writing the same things every time. So, for example, I suggest formatting the captions as:
   6. Figure 6. Same as in Figure 4, but for (a) MAM, (b) March, (c) April, and (d) May.
   7. Figure 11. Same as in Figure 5, but for SON.
4. This comment is a recommendation for improving the visual appearance and accessibility of the figures: The trend line colors in the Sen’s slope plots (Figures 3,5,7,9,11) should be colorblind friendly. I suggest using a colorblind friendly palette for the plots lines colors. For example, the yellow color in figure 11 is too bright and not easy to see.

Another recommendation is to add a table that summarizes the findings. For example, a table with the rows corresponding to the seasons or months (could be like annual, DFJ, MAM, JJA, SON) and the columns to the different regions (like Artic Ocea, Southern Ocean, Northern China, Tropical East Pacific, etc.) filled with the increasing/decreasing numerical values (mm/day/yr). I believe this table will significantly enhance clarity and reduce the heavy numeric text.

Line 156: “statically” - > “statistically”

Line 231: NAO - > North Atlantic Oscillation

Conclusions

The Conclusions section clearly summarizes the study’s objectives, methodology and key findings. Integration with literature, mechanism and limitations are stated which strengthens the scientific value of the study. A future work text with clear ideas is also included.

Line 268: “the the” - > “the”

Author Response

Please check the uploaded document. 

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

General comments

The manuscript uses NASA’s Integrated Multi-satellite Retrievals for GPM (IMERG) Version 7 global high-resolution precipitation analysis to study the annual, seasonal and monthly trends of precipitation in the years 1998-2024. It documents a highly heteregenous distribution of precipitation trends during this period, including a statistically significant wettening of the Arctic Ocean and high-latitude Southern Ocean as well as eastern China, combined with significant drying in the southeastern tropical Pacific and parts of South America and South-to-Central Africa.

This trend analysis is to my knowledge novel in the sense that the same data set has not been used for this purpose before. The research methods are straightforward but adequate. The same applies to the way in which the results are presented, although I do have a few suggestions for improvement.

The only part of the methodology that I am somewhat reserved to is the choice to only calculate trends for those areas where they are statistically significant.  Statistical significance only means that the observed trend is large compared to the expected magnitude of random error. It does not mean that the significant trends would be, in an absolute sense, closer to the true underlying forced trend than the insignificant ones. In fact, the reverse may happen, because the largest and apparently most significant trends are often observed in areas where the observed trend is accidentally amplified by an unusually large contribution from internal variability. This is especially true for variables like precipitation, for which the signal-to-noise ratio between the forced trend and internal variability is relatively low.

Thus, ideally, the areas for which the trends are reported should be chosen a priori, for example using the same regional division as in the latest IPCC report. However, I see that this choice may be difficult to change at this stage of the research.

Detailed comments on the presentation of the results

1. Section 3. In addition to the maps and regional time series, it would be useful to also show the time series and the trend in the global mean precipitation. This would give the readers some idea of the temporal homogeneity of the IMERG product, which might be otherwise difficult to assess. If the trend in the global precipitation is anything else than a minor increase of ca. 0-1 mm per year, there might be a problem.
2. Please outline in Figures 2, 4, 6, 8 and 10 the regions for which the time series and trends are shown in Figures 3, 5, 7, 9 and 11. Currently, their definition is too diffuse.
3. Figures 4, 6, 8 and 10. Add the seasons and month directly in the headings of the panels. For example, for Figure 4: (a) DJF, (b) December, (c) January and (d) February.

Other minor comments

1. L62-63. You should probably refer to [6] rather than [7]. More importantly, the sentence is misleading. [6] found that the 1 % largest (in area) precipitation events contribute 80 % to the total precipitation. The largest events are not the same as the most intense events.
2. L153-155. Only five clusters are listed.
3. Arctic with capital A.
4. L161-162. "decades earlier and at a lower level of global warming": compared with what?
5. In the Northern Hemisphere winter (DJF), four main clusters dominate ...
6. The results for the Northern Hemisphere spring (MAM) show ...
7. The results for the Northern Hemisphere summer (JJA) show ...
8. For the Northern Hemisphere autumn (SON)?
9. L276-277. were calculated for areas with statistically significant trends

Author Response

Please check the uploaded document. 

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

My questions have been adequately addressed. No further comments.