CommunityImagegithub.com

research-figures-maps

Plan, script and check publication figures and maps for watershed, water-quality, agricultural and food-system papers — hydrographs with precipitation, flow-duration and load-duration curves, observed-versus-simulated plots, BMP and scenario comparisons with uncertainty, study-area and result maps (scale bar, north arrow, CRS, data credits), colour-blind-safe palettes, journal sizes and captions. Use whenever the user asks to make, fix, critique or redesign a figure, plot model or monitoring results for a manuscript, build a study-area map, choose colours or a chart type, or write figure captions, even for a single plot.

What is research-figures-maps?

research-figures-maps is a Claude Code agent skill that plan, script and check publication figures and maps for watershed, water-quality, agricultural and food-system papers — hydrographs with precipitation, flow-duration and load-duration curves, observed-versus-simulated plots, BMP and scenario comparisons with uncertainty, study-area and result maps (scale bar, north arrow, CRS, data credits), colour-blind-safe palettes, journal sizes and captions. Use whenever the user asks to make, fix, critique or redesign a figure, plot model or monitoring results for a manuscript, build a study-area map, choose colours or a chart type, or write figure captions, even for a single plot.

Works with~Claude Code~Codex CLI~Cursor
npx skills add https://github.com/bdzion/claude_skills_z/tree/HEAD/research-figures-maps

Installed? Explore more Image skills: steipete/songsee, affaan-m/frontend-design, mattpocock/improve-codebase-architecture · View all 6 →

Ask in your favorite AI

Open a new chat with this agent skill pre-loaded.

Documentation

Research figures and maps

A figure earns its place when it shows something a table cannot (a pattern over time or flow, a distribution, a spatial arrangement) and reads at print size in grey scale. Every scripted figure reads its data from files, so it cannot drift from the numbers in the text.

Workflow

  1. Decide what the figure must show in one sentence and check that a table would not do it better. Note the target journal and its column widths.
  2. Pick the form from references/figure-guide.md: variable types, axes, transformations (log for flow and loads), and how uncertainty is shown.
  3. Script it with the house style module scripts/zfig.py: use_style(), figure("single"|"onehalf"|"double", height_mm), and recipes hydrograph_with_rain, flow_duration, load_duration, obs_vs_sim, scenario_bars. One script per figure (figN_short_name.py), data read by explicit path, units in every axis label, and an assert for each number the text quotes from the figure. Run python scripts/zfig.py once to self-test in a new environment.
  4. Maps: follow the map checklist in the guide (CRS stated, scale bar, north arrow when orientation is not obvious, inset locator, legend with units, data sources and licence credits in the caption). For final layouts QGIS or ArcGIS Pro is usually quicker than matplotlib; script the data preparation either way.
  5. Look at the rendered file at its final size before calling it finished: labels clipped or overlapping, legend covering data, ticks unreadable, colours collapsing in grey scale or for colour-vision deficiency.
  6. Caption: what is shown, then how to read it (symbols, lines, shading, units, period, n). Keep interpretation in the text.

Defaults that matter

  • Okabe–Ito qualitative palette (built into zfig); viridis, cividis or Crameri scientific colour maps for continuous data; diverging maps only around a meaningful zero (e.g. change from baseline). No rainbow or jet (Crameri et al. 2020).
  • 8 pt text at final size, 7 pt minimum; line widths 0.6–1.0 pt.
  • Observed data in black or dark grey and simulations in colour, consistently across figures.
  • Log axes for flow, loads and concentrations spanning orders of magnitude; say so in the label.
  • Show uncertainty (95PPU bands, ranges across scenarios or climate members, confidence intervals); a bar without its spread overstates certainty.
  • Do not invent data, coordinates or labels; if data are missing, write the script with a clear NotImplementedError and list what is needed.

Python first and commented; state units, CRS and time step in each figure script's docstring.

Individual skills in this repo

This repo contains 1 individual skill — each has its own dedicated page.

Related Skills