the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 23 Jul 2026
MSDT v1.0: An Open-source Multi-method Snow Drought Toolbox
Abstract. Snow drought, characterized by anomalously low snow water equivalent (SWE) relative to climatological conditions, has become an increasingly important cryospheric hazard with substantial environmental and socio-economic impacts. Although numerous snow drought identification methods have been developed, existing implementations are typically fragmented into independent scripts and region-specific workflows, hindering methodological consistency, reproducibility, and systematic intercomparison. Here we present the Multi-method Snow Drought Toolbox (MSDT), an open-source Python software package that provides an end-to-end workflow for snow drought research, encompassing SWE data acquisition, multi-method identification, data visualization, and comparative analysis through an intuitive graphical user interface. MSDT integrates seven widely adopted snow drought identification methods spanning yearly, monthly, and daily temporal scales, with flexible and adjustable parameters that enable users to tailor analyses to different climatic regions and research objectives. The toolbox streamlines data access by supporting automated downloading of state-of-the-art SWE datasets together with a curated catalog of commonly used products, and further supports interactive geospatial visualization for data exploration and methodological comparison. In addition, the toolbox adopts a memory- and performance-efficient architecture that supports large spatiotemporal datasets while maintaining responsive snow drought computing. Its modular architecture enables straightforward integration of additional datasets and identification algorithms. We demonstrate the capability of MSDT using ERA5-Land SWE data to reproduce global snow drought climatology, temporal variability, and representative historical events documented in previous studies. The results show that MSDT consistently reproduces published snow drought characteristics while substantially improving computational efficiency, workflow reproducibility, and methodological comparability. By providing a standardized, extensible, and user-friendly platform, MSDT lowers technical barriers and facilitates reproducible snow drought monitoring, model evaluation, and impact assessment.
- Preprint
(2566 KB) - Metadata XML
-
Supplement
(370 KB) - BibTeX
- EndNote
Status: open (until 17 Sep 2026)
- RC1: 'Comment on egusphere-2026-4290', Anonymous Referee #1, 28 Jul 2026 reply
-
RC2: 'Comment on egusphere-2026-4290', Anonymous Referee #2, 29 Jul 2026
reply
This manuscript presents the Multi-method Snow Drought Toolbox (MSDT), an open-source Python framework integrating multiple snow drought identification methods across yearly, monthly, and daily temporal scales. The authors aim to address the fragmentation of existing snow drought analysis workflows by providing a unified platform for SWE data acquisition, snow drought detection, visualization, and standardized output generation.
The development of such a toolbox is timely and potentially valuable for the cryosphere research community. The manuscript demonstrates the capability of MSDT through global-scale analyses using ERA5-Land SWE and several representative snow drought events. The modular design and integration of multiple approaches represent an important step toward improving reproducibility in snow drought studies. Several concerns should be addressed before publication. I recommend a Moderate Revision.
Major Concern 1: The algorithm implementation requires greater transparency
The manuscript integrates seven snow drought identification methods. However, the technical description is insufficient for independent reproduction. Important missing information includes detailed algorithm workflow, input/output structure, handling of missing values, threshold implementation, etc.
For example, SWEI allows multiple probability distributions, but the criteria for selecting or rejecting distributions are unclear. The authors should provide algorithm pseudocode, flowcharts for individual methods, example parameter files if applicable, especially PDM and SWEI.
Major Concern 2: The software contribution lacks sufficient quantitative evaluation of computational performance
The manuscript highlights several computational improvements, including lazy NetCDF loading, adaptive parallelization, float32 optimization, and compressed NetCDF output. However, these improvements are described qualitatively. The authors should provide quantitative benchmarking which describing the hardware configuration, dataset size, runtime, peak memory consumption, and comparison with traditional scripts. These would significantly strengthen the software contribution.
Minor Concerns
- The last two paragraphs of the Introduction are somewhat repetitive and should be revised to avoid redundancy.
- The manuscript should clarify whether snow drought intensity has consistent meaning among different methods.
- The use of ERA5-Land SWE as the primary demonstration dataset requires additional uncertainty discussion.
- The snow-covered area filtering approach needs more justification. The choices of 5% MODIS threshold and 75% SWE occurrence threshold appear empirical.
- The manuscript allows different probability distributions for SWEI but does not discuss their influence on results.
- The role of temperature, radiation, and rain–snow transition in snow drought development should be discussed more explicitly.
- The manuscript should clarify whether PDM can distinguish warm snow drought and dry snow drought.
- Figure 1 is informative but the text size is too small for publication.
- Parallel computing strategy requires more technical details.
- The NetCDF output structure should be documented according to CF conventions
- Table 2 should include data types and units for each output variable.
- The visualization module should support modern mapping libraries instead of relying only on Basemap, which is becoming outdated.
- The manuscript should provide example workflows in the Supplementary Material.
- Claims regarding “standard desktop computers” are too general without hardware specification.
Citation: https://doi.org/10.5194/egusphere-2026-4290-RC2 -
RC3: 'Comment on egusphere-2026-4290', Anonymous Referee #3, 31 Aug 2026
reply
Reviewer Comments
Review of: MSDT v1.0: An Open-source Multi-method Snow Drought Toolbox (egusphere-2026-4290)General Comments
I read this manuscript with great interest and genuine enthusiasm. The snow drought research community has long needed a unified computational platform, and MSDT represents exactly the kind of foundational infrastructure that facilitates reproducible, intercomparable science. The authors have done commendable work in harmonizing seven different snow drought identification methods across multiple temporal scales, packaging them with automated data access and interactive visualization tools, and demonstrating their application through well-chosen and geographically diverse case studies. The modular architecture and standardized NetCDF output are particularly thoughtful design choices that will serve the community well.
That said, I believe the manuscript can be substantially strengthened in several areas. My comments below are intended as constructive suggestions to help the authors realize the full potential of this contribution.Major Recommendations
1. Add a quantitative method intercomparison framework to demonstrate MSDT's unique value proposition.
One of the most exciting promises of MSDT is that it enables systematic comparison among snow drought definitions. The current manuscript demonstrates this qualitatively (Figs. 3-11) but stops short of providing quantitative metrics that would help users understand when and why methods diverge. I encourage the authors to consider adding:
• A spatial agreement map (e.g., the fraction of methods that agree on drought classification at each grid cell) for a representative year, which would immediately reveal regions of high vs. low methodological consensus.
• A pairwise confusion matrix or Cohen kappa among methods for drought/no-drought classification, perhaps stratified by climate zone (e.g., maritime, continental, alpine, Arctic).
• Timeseries of the "ensemble spread" in drought area across methods, which would quantify how much the choice of method affects the global drought trend narrative.
These additions would transform Sect. 3 from a demonstration of functionality into a genuine scientific contribution - a quantitative assessment of methodological uncertainty in snow drought research. This is where MSDT can truly add unique value beyond being a software package.
2. Provide quantitative software performance metrics to substantiate efficiency claims.
Section 2.3.5 describes an impressive optimization strategy, and I appreciate the care taken with lazy loading, float32 precision, adaptive parallelism, and chunked output. However, the evaluation remains qualitative. I suggest adding a brief benchmarking section or table:
• Test platform specifications (CPU model, RAM, OS, Python version).
• For a representative global ERA5-Land dataset (1980-2024, 0.25 deg, snow-covered domain): wall-clock time and peak RAM for each of the seven methods.
• Storage comparison: NetCDF4 compressed vs. uncompressed file sizes for a standard output set.
These benchmarks would transform the optimization section from a description of intent into a documented achievement, greatly strengthening the software contribution.
3. Situate MSDT more explicitly in the broader drought informatics landscape.
The Introduction appropriately mentions meteorological and hydrological drought indices (Hao et al., 2017; Hayes et al., 2011) as models for MSDT. I suggest expanding this comparison in the Discussion by drawing concrete parallels with successful drought software packages (e.g., SPEI R package, scPDSI, xclim). What design lessons from these mature packages informed MSDT? What gaps remain? This would help readers understand MSDT not as an isolated effort but as part of a broader movement toward standardized, community-vetted climate hazard software.Specific Suggestions
Introduction (L64-67): The paragraph on existing methods is effective. A small suggestion: explicitly tabulate which method combinations have never been compared in a single study before MSDT. This would sharpen the novelty claim by showing the gap MSDT fills.
L134-171: SWEI/PDM descriptions are vague. How is probability distribution selected? What smoothing is applied? These details matter for reproducibility.
Section 2.2 (L172-178): The snow cover masking procedure is important but described too briefly. Please add: (a) MODIS product version and collection number; (b) regridding method to 0.25 degrees; (c) justification for the 5% and 75% thresholds with sensitivity tests or supporting citations.
Section 2.3.3 (Visualization): Basemap is deprecated in favor of Cartopy. Please justify this dependency choice for long-term maintainability, or migrate to Cartopy.
Sect. 2.3.5: Adaptive parallelism description is hand-waving. How is available memory detected programmatically? What is the exact partitioning heuristic? Provide pseudo-code.
Section 3.1 (L296-330): 22-year split comparison is too crude. Suggest using Theil-Sen slopes with Mann-Kendall significance testing to provide rigorous, comparable trend characterization.
Discussion (L415-443): Add a paragraph on the limitations of SWE-only approaches vs. multi-variable drought indicators (e.g., incorporating temperature for warm vs. dry snow drought discrimination). This would guide users and set the stage for planned future extensions.
References: Multiple encoding errors in accented names. Several missing DOIs. Journal name abbreviations are inconsistent (full names vs. abbreviated).Summary and Recommendation
MSDT addresses a genuine and well-motivated need in the snow hydrology community. The software design is thoughtful, the multi-method integration is valuable, and the demonstration cases are well-chosen. I am optimistic about this contribution for snow drought research.
My primary recommendation is that the authors consider the major additions suggested above—particularly quantitative method intercomparison and software performance benchmarks — which would elevate the manuscript from a solid software announcement into a contribution that simultaneously advances both snow drought science and community cyberinfrastructure.Recommendation: Major Revision. I look forward to seeing a strengthened version of this manuscript.
Citation: https://doi.org/10.5194/egusphere-2026-4290-RC3
Data sets
MSDT_Input_Output Xiangfei Li and Jinzhi Tian https://zenodo.org/records/21396300
Model code and software
MSDT_Software Xiangfei Li and Jinzhi Tian https://zenodo.org/records/21396772
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 96 | 34 | 14 | 144 | 22 | 10 | 13 |
- HTML: 96
- PDF: 34
- XML: 14
- Total: 144
- Supplement: 22
- BibTeX: 10
- EndNote: 13
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Overall Evaluation
This manuscript develops a fully integrated open-source Python toolbox (MSDT) for standardized snow drought assessment across yearly, monthly, and daily temporal scales. The authors integrate seven widely adopted snow drought identification algorithms, embed automated SWE data downloading, interactive geospatial visualization, and standardized NetCDF output within a graphical user interface. Global ERA5-Land datasets and three well-documented high-impact snow drought events are used to validate the toolbox performance. The work addresses a clear community gap: fragmented, non-reproducible custom scripts dominate current snow drought research, while no unified platform supports multi-method comparison. The modular software architecture, end-to-end computational optimization, and fully open code/datasets are significant strengths. Overall, this is a valuable methodological contribution to cryosphere and hydrological drought research, but one core issues about the computational performance require substantial revisions, alongside some minor textual, graphical and structural adjustments before acceptance.
Major Concern：
Computational performance benchmarks lack standardized hardware reference and scalability tests.
The manuscript introduces multi-layer computational optimization strategies for MSDT, but performance descriptions remain descriptive without standardized benchmarks.
1. No unified hardware environment is specified for runtime comparisons across yearly, monthly SWEI, and daily PDM modules. Readers cannot replicate or reference the reported efficiency gains.
2. Scalability tests are missing: the paper does not evaluate how processing time and memory consumption scale with spatial resolution (e.g., 0.25°, 0.1°, 0.01°) or temporal length of input datasets.
3. Revision request: Add a standardized performance benchmark table with clear hardware specifications, record runtime and peak memory usage for all three temporal modules under multiple spatial/temporal input configurations, and provide practical hardware recommendations for users running large-domain high-resolution analysis.
Minor Concerns：
1. Abstract & Introduction redundancy: The opening background repeatedly restates identical socio-environmental impacts of snow drought across two paragraphs. Condense repetitive literature summaries and tighten the narrative to highlight the core research gap (lack of unified multi-method toolbox) earlier.
2. Typographical errors in figure labels: Multiple spelling mistakes appear in map captions, e.g., “Muti-year mean daily snow drought intensity” (Figure 5a) should be “Multi-year”. The authors must systematically proofread all figure titles, axis labels and legends for spelling/grammar errors.
3. Ambiguous snow mask preprocessing workflow: Section 2.2 describes snow persistence zone filtering using MODIS and ERA5-Land thresholds, but does not clarify whether users can upload custom snow cover masks into MSDT. Explicitly state this customizable function in the text.
4. Figure annotation clarity: Circled anomalous regions in Figure 6g–h lack clear explanatory text linking peak SWE timing shifts to observed regional storm events. Expand the figure caption to connect visual anomalies to the cited National Weather Service records.
5. Hydrological year setting explanation: The tool automatically assigns winter periods for Northern/Southern Hemispheres, but the manuscript does not explain how users can manually override the default hydrological calendar. Add a short note on custom winter period configuration in Section 2.3.2.
6. Missing severity classification metadata explanation: Table 2 lists output variables including `intensity_flag` (drought severity level), yet the paper does not detail the default severity tiers for yearly/monthly/daily methods. Add a concise table summarizing severity thresholds for all seven algorithms.
7. Reference formatting inconsistency: Several journal article references use inconsistent abbreviation styles for journal names, and page number ranges are formatted unevenly. Standardize all citations.
8. The phrase “independent yet interconnected modules” should be corrected to “independent but interconnected modules”.
9. The manuscript alternates between “snow drought” and “SD”. Please standardize.
Final Recommendation
The proposed MSDT toolbox delivers substantial value to the cryosphere and hydrology community, and the global/regional case studies effectively demonstrate its core functionality. The manuscript is promising but requires full addressing of all listed concerns. I recommend a Moderate Revision, and look forward a second round of evaluation.