---
name: academic-writing-style
description: Write or revise scientific text in the user's own academic style, learned from their published climate papers (drought, heatwaves, ENSO teleconnections, regional climate models / bias correction over Vietnam and Southeast Asia). Use whenever the user asks to draft, rewrite, polish, or review any part of a research paper, abstract, proposal, report, thesis chapter, or reviewer response "in my style", or asks for scientific writing in climatology, meteorology, hydro-meteorology or related geoscience.
---

# The user's academic writing style

This skill comes from four first-author or corresponding-author papers. Two are in *Int. J. Climatol.* (2019 and 2024), one is in *J. Climate* (2022) and one is in *Climatic Change* (2025). The voice is that of a senior climate scientist. It is clear, methodical, evidence-led and cautious. The prose is plain and has few rhetorical flourishes. Each claim is tied to a figure, a table, a number or a citation.

Before drafting a section, read `references/section-templates.md` for that section's structure and model sentences. Use `references/phrasebank.md` for transitions and stock phrases.

## Core voice

1. **Objective and evidence-first.** Every result sentence points to its evidence ("Figure 4 shows…", "(Table 2)", "(Fig. 5d)") or gives a number. Do not make an unsupported claim.
2. **"We" plus the passive.** Use "we" for choices and actions ("we investigate", "we applied", "we identified"). Use the passive for procedures ("PET is calculated using the Thornthwaite method", "16 climate indices were analysed"). Never use "I".
3. **Cautious interpretation.** Report observations firmly. Hedge the physical explanations: *suggesting, indicating, implying, could be attributed to, might be related to, it seems that, appears to, likely*. Save "confirms" and "demonstrates" for results that are robust and statistically significant.
4. **Physical reasoning after each result.** After a pattern, explain the mechanism. Typical mechanisms are monsoon timing, ENSO and the Walker circulation, topography such as the Foehn effect on the lee side of the Truong Son range, latitude, altitude, rainy-season timing and moisture transport. Use "This could be attributed to…", "This is to be expected, due to…" or "as a result of…".
5. **Paired opposites in parentheses.** This is a signature device: *"positive (negative) values of EOF1 are associated with positive (negative) values of PDSI"*, *"drought (wetness) conditions … during El Niño (La Niña) years"*, *"increasing (decreasing) trends"*, *"R2 (R1)"*. Use it whenever two symmetric cases can share one sentence.
6. **Geographic structure by sub-region.** Organise results by region codes, such as R1–R7 for Vietnam's seven climatic sub-regions or R1–R4 for SEA clusters. Contrast them, for example north vs. central vs. south, or mainland Indochina vs. Maritime Continent. Name each region in full with its code at first use: "Northwest (R1), Northeast (R2), Red River Delta (R3), North Central (R4), South Central (R5), Central Highlands (R6), and the South (R7)".
7. **Quantified ranges.** Prefer specific ranges and values: "in the range of 15%–20%", "ranging from +0.32 to +0.71 days", "accounting for 57% and 59%, respectively", "|R| mostly ranging from 0.33 to 0.45". Use "respectively" to map lists of values onto lists of items.
8. **Relevance to applications.** Abstracts and conclusions end with practical value. Typical endings point to early prediction or a seasonal forecasting or early-warning system, drought or heatwave management, agriculture and water resources, or climate projections. Add a line on future work after that.

## Mechanics

- **Abbreviations.** Write the full term at first use with the acronym in parentheses, then use the acronym throughout: "Standardized Precipitation Evapotranspiration Index (SPEI)", "climate indices (hereafter referred to as CIs)". Define a characteristic and its symbol together: drought frequency (DF), duration (DD), severity (DS), spatial extent (SE); HWN, HWL, HWD, HWM.
- **Ranges and periods.** Use an en dash: "1980–2014", "R1–R3", "April–September". Phrase periods as "for the period 1960–2019" or "during the period 1980–2020". Write decades as "1990–99" in AMS style. Write ENSO events as "1997/98" or "2015/16".
- **Seasons.** Use month initials, defined at first use: DJF, MAM, JJA, SON, AMJ, ONDJF.
- **Significance.** "at the 5% significance level", "satisfy the 95% confidence level of the t-test", "Black circles indicate p-value < .05", "statistically significant based on the North test".
- **Spelling.** Follow the target journal: British for RMetS/Wiley (*analyse, behaviour, whilst, amongst, colour*) and American for AMS/Springer (*analyze, behavior*). Ask if the journal is unknown. Default to British for *Int. J. Climatol.*
- **Citations.** Use author–year, often in clusters (*Mishra and Singh, 2010; Le et al., 2019*). Introduce examples with "e.g.," inside parentheses. Cite the user's group's own earlier work where relevant (Le et al. 2019; Phan-Van et al. 2018, 2022; Pham-Thanh et al. 2019; Nguyen-Ngoc-Bich et al. 2021, 2022).
- **Cross-references.** Use "(see Section 2.4)", "(see details in Figs. S3 and S4)", "(not shown)", "(Figure S6 in the Supplementary)".
- **Paragraphs.** Paragraphs are medium-long, 5–12 sentences, and each develops one figure or idea. They open with a topic sentence and often close with a synthesis sentence: "Overall, …", "In general, …", "In summary, …".
- **Equations.** Introduce them with "can be briefly shown as follows:" or "is calculated as follows:". After each one, add "where X is …, Y is …, and k is the number of …".
- **Lists.** Number definitions of characteristics and conclusion findings: (1)… (2)… or 1) 2) in AMS style. Use (i), (ii) inline.
- **Figure captions.** Start with a noun phrase: "Spatial distribution of…", "Time series of…", "Composite maps of…". Label panels: "(a) … (b) …". Write repeated captions as "As in Fig. 7, but for SPEI and SST." or "Same as Fig. 2, but for …".

## Quality bar

- Keep the voice, but write clean, grammatical English. Fix the small slips that appear in the source papers, such as subject–verb agreement, missing articles, "following by" for "followed by", and "principle" for "principal" component.
- Do not invent numbers, citations or results. If the user has not supplied a value, insert a placeholder such as `[XX%]` or `[citation]` and flag it.
- Do not use hype words such as *groundbreaking, paradigm-shifting, delve, crucial tapestry, unprecedented insights*. The strongest novelty claim the user makes is "To the best of our knowledge, this study is the first…" or "This is the first time … has been comprehensively assessed…".
- Avoid em-dash-heavy, listicle-style or bullet-heavy prose in the body. The papers use continuous paragraphs. The only bullets are numbered definitions and numbered conclusions.

## Workflow

1. Identify the section (abstract, introduction, data and methods, results, discussion, conclusions, captions, response to reviewers) and the target journal.
2. Open `references/section-templates.md` and follow that section's sequence of moves.
3. Draft with the user's facts. Mark any missing facts as placeholders.
4. Self-check against the Core voice list: evidence-linked, physically interpreted, hedged correctly, paired-opposite device used where natural, ranges quantified, acronyms defined once.
5. When revising the user's text, keep their content and terminology and change only what the style or the grammar requires.