Data visualization tools for communicating food security research

Food security research generates enormous volumes of information: yield figures from remote sensing, household consumption surveys, climate projections, and trade statistics that arrive in incompatible spreadsheets. The IASSIST 2017 gathering in Lawrence brought practitioners together to wrestle with how researchers can turn those raw numbers into visual stories that decision-makers actually use. Across the Pacific, Australian researchers face comparable pressures, working in a country where productive land stretches from the tropical orchards of the Atherton Tablelands to the broadacre wheat belts of Western Australia, and where seasonal rainfall variability can transform a record harvest into a feed shortage within twelve months.

The Australian science community has invested heavily in visualisation capacity over the past decade. CSIRO's Data61 unit, the Australian Bureau of Agricultural and Resource Economics and Sciences (ABARES), and the Grains Research and Development Corporation all publish interactive dashboards that summarise commodity outlooks, drought conditions, and supply chain risks. University research groups in Melbourne, Sydney, and Adelaide have developed open-source R and Python packages that translate agricultural census data into choropleths, time-series ribbons, and small-multiple panels. This article walks through the practical toolset that communicators in Australia are using right now, from desktop platforms to browser-native libraries.

Picking the right tool often comes down to the audience. A parliamentary briefing in Canberra needs a static PDF chart that prints cleanly on A4. A grower-facing workshop in regional New South Wales benefits from an interactive dashboard that runs on a slow rural internet connection. Researchers writing journal articles in English-speaking outlets need figures that meet strict accessibility standards and reproduce cleanly in print. The next sections examine the leading visualisation platforms, mapping extensions, and storytelling formats, and close with a side-by-side comparison of six tools commonly seen in Australian food security work.

From spreadsheets to visual insight

The first decision any research team faces is whether to stay inside a familiar spreadsheet environment or move to a dedicated visualisation platform. Excel and Google Sheets remain the default starting place for many ABARES analysts and university postgraduate students. Their charting menus produce acceptable bar charts, line graphs, and scatter plots, but they falter when the output must layer categorical groupings, animate time series, or apply consistent brand colours across dozens of figures. Pivot tables help summarise tonnage by state, yet they rarely produce the publication-ready artefacts that journal reviewers expect.

Power BI and Tableau address those shortcomings by treating visualisation as a layered canvas rather than a one-off chart. Power BI has gained ground because it ships with Microsoft 365 licences that most Australian government departments already hold, and because its DAX formula language makes it straightforward to model relationships between production tables, weather records, and trade flows. Tableau's strength lies in its drag interface and its vibrant user community, including a strong Sydney group that runs regular show-and-tell sessions for analysts from NSW Health, the Department of Agriculture, and several state universities. Researchers who already write in Python or R often prefer to skip both platforms and use code-native libraries that integrate directly with their analysis pipelines.

Code-native visualisation offers reproducibility, which matters for grant reporting. ggplot2 in R, Plotly in Python, and Observable in JavaScript all let researchers build a figure once and re-render it whenever the underlying data changes. The trade-off is a steeper learning curve and the need for version control literacy, which is why many teams adopt a two-speed approach: prototype quickly in Tableau or Power BI for stakeholder feedback, then rebuild the final version in R or Python for publication. Teams travelling for stakeholder workshops or partner meetings often sort out hotel arrangements well in advance, since major research events in Australian cities book out months ahead.

Mapping food production across the continent

Geospatial visualisation deserves special attention in any Australian food security conversation, because distance and climate variability make maps the most efficient way to communicate where production happens. The Bureau of Meteorology's climate datasets pair naturally with ABARES production statistics, and tools like QGIS let researchers overlay rainfall deciles, soil moisture, and crop type polygons on a single canvas. Researchers in Queensland have used this combination to visualise how the 2017-2019 drought reduced sorghum yields across the Darling Downs, communicating a complex climate-production story in a single static figure.

ArcGIS Online and the open-source kepler.gl extend this capability into the browser. Both platforms ingest GeoJSON files directly from a researcher's analysis pipeline and produce interactive maps that rotate, zoom, and respond to layer toggles. The Australian Institute of Food Science and Technology has promoted the use of such tools for traceability work, particularly around horticultural exports from the Sunraysia and Riverland regions. Researchers producing figures for ABC news stories or the Conversation often build their maps in kepler.gl, then export static snapshots that retain the same colour logic as the interactive version.

Map design conventions matter in this domain. Researchers should adopt consistent colour ramps for rainfall deciles, follow the Government of Australia's style guide for typography, and avoid the green-to-red spectrum that causes problems for colour-blind readers. Embedding a north arrow, scale bar, and data source attribution inside the map frame signals rigour and helps the figure survive re-use by other journalists or policy analysts. Maps of remote regions, from the Top End to the Pilbara, also need a sensible choice of projection so that distance and area read correctly for an Australian audience.

Interactive dashboards for stakeholder engagement

Dashboards extend maps and tables into a navigable workspace where stakeholders can interrogate the data themselves. The most successful Australian examples combine a clear narrative with exploratory freedom. ABARES' Agricultural Commodities dashboard leads with a written outlook, then layers interactive charts underneath that readers can filter by commodity, state, and time horizon. The NSW Department of Primary Industries runs a Seasonal Conditions dashboard of its own, covering livestock, cropping, and horticulture indicators, and it follows the same pattern of leading prose before interactive elements.

R Shiny remains the workhorse for academic and government research teams that already work in the language. It integrates tightly with the tidyverse, deploys to shinyapps.io with a couple of commands, and supports custom CSS for branding. Python alternatives include Streamlit and Dash by Plotly, both of which have matured into reliable platforms for production dashboards. Streamlit's appeal is its low-code layout: a researcher can spin up a filterable dashboard in an afternoon, and the framework handles state management automatically. Dash offers finer control over callbacks and is the preferred choice for teams that need authentication, complex workflows, or integration with enterprise data warehouses.

The biggest practical challenge for dashboards in rural Australia is bandwidth. Many growers and extension officers access the internet through 4G networks on the edge of mobile coverage, and dashboard pages that load megabytes of JavaScript will frustrate them. Researchers should design dashboards that ship a minimal initial payload, lazy-load heavy visualisations, and offer a downloadable PDF or PNG summary for offline reading. That offline export doubles as a print-friendly summary for committee meetings where tablets are not welcome.

Choosing the right tool for the message

The tool has to fit the audience, the data shape, and the medium. A static figure for a peer-reviewed journal demands reproducible code, accessible colour palettes, and high-resolution output at 300 dpi. PowerPoint decks for a workshop in a regional centre tolerate lower resolution but require figures that read clearly from the back row. Web dashboards for grower associations need to function on a six-year-old tablet running Android.

Tableau Public offers a generous free publishing tier and produces interactive visualisations that embed cleanly in news articles, but it ties the researcher to Tableau's ecosystem. Power BI Pro is more affordable for Australian organisations that already pay for Microsoft licensing, and its integration with Excel makes it the path of least resistance for analysts who already build pivot tables. R Markdown and Quarto produce publication-quality PDFs and HTML reports from a single source, which suits university research groups writing reports for the Rural Research and Development Corporations. Observable notebooks offer a collaborative, JavaScript-native environment that suits data journalism teams, particularly those producing explainer pieces for the Conversation or for ABC's specialist reporting teams.

The choice also depends on who will maintain the visualisation after the researcher moves on. Open-source code in a public repository survives staff turnover and survives grant cycles. Proprietary dashboards require ongoing licence fees and an analyst who knows the platform. Australian research groups funded through the Australian Research Council often face this dilemma halfway through a project, and the safer path is usually to invest a little more time up front to build the visualisation in code, even if the prototype was faster to assemble in a drag-and-drop tool.

Workflows, comparison, and recommendations

The workflow that works best in practice starts in a version-controlled repository on GitHub or GitLab, with raw data stored separately from analysis code. Researchers develop the analysis in R or Python, commit regularly, and use renv or venv to pin package versions. They then publish figures through a layered stack: static publication figures exported at print resolution, interactive dashboards built with Shiny or Dash for stakeholder engagement, and embeddable charts built with Plotly or D3 for news stories and policy briefs. Cold-chain logistics data, particularly for remote communities that face food preservation challenges across the Top End and Pilbara, often surfaces through partnerships with research networks such as Thermo International, which maintain open datasets on refrigeration performance in tropical climates.

Tool	Best for	Skill level	Cost	Reproducibility	Offline support
Tableau Public	Interactive news graphics	Low–medium	Free (public)	Limited	Limited
Power BI Pro	Government and enterprise reports	Low–medium	Per user/year	Medium	Medium
R Shiny + ggplot2	Academic dashboards	Medium–high	Open source	Excellent	Strong
Streamlit + Plotly	Rapid Python prototyping	Medium	Open source	Excellent	Strong
QGIS	Geospatial analysis	Medium	Open source	Excellent	Strong
Datawrapper	Quick embeddable charts	Low	Free tier	Good	Strong

Each of these tools can produce credible food security visualisations when paired with good data and clear narrative. The wrong choice is rarely fatal; researchers can always re-cut their work in a different platform once the underlying analysis is sound.

A few habits consistently improve the reach of food security visualisations across Australian contexts.

• Begin the figure in the same repository as the analysis, and store raw data separately from figures and rendered output.
• Choose the colour palette before drafting the first chart, and check it against a deuteranopia and protanopia simulator.
• Lead with a written interpretation above every dashboard, so readers do not need to interpret the chart unaided.
• Produce both a high-resolution static export for print and a lightweight interactive version for the web.
• Test the dashboard on a mid-range Android device over a throttled 4G connection to simulate rural Australian users.
• Attribute every figure with the data source, the analysis date, and a contact email, so journalists and policy analysts can follow up.
• Document the visualisation in a README that explains the basis, the limitations, and how to reproduce the figure from the raw data.