Field Practice →
The seven capture methods, the pre-fieldwork timeline, and the two sensory-mapping experiments.
Data & Evidence →
Real data preview, the polar/desert satellite timelapse, the dataset framework, and references.
Production Diary →
How to read these sensory mappings
Every mapping on this page follows the same chain: satellite measurement → value → artistic rule → sensory form. The sensory form is one possible interpretation of the value, built through artistic and technical decisions — not the value's natural or inevitable meaning. Making that chain visible is the point, not a limitation: it shows that environmental knowledge is constructed through instruments and practices (Gabrys, 2016), rather than delivered untouched from nature.
Smell needs particular care. Satellites do not measure scent, ice age, or the smell intensity of desert dust. Where this site pairs a dataset with smell, it means a deliberate artistic association — a scent captured at the field site, an aromatic plant associated with the location, a captured airborne compound, a mineral/marine/soil-inspired composition, or an archival scent composed for a particular environmental moment. A satellite reading may control that scent's release timing or intensity, but never its content.
No single sensory form explains the dataset. Sound may communicate change over time; touch may communicate contrast; light may create atmosphere; movement may communicate scale or transition; smell may evoke place, memory and environmental character. The plurality matters — it prevents any one artistic reading from claiming to be the only correct meaning of the data. Visitors are not literally feeling the satellite data; they are encountering a sensory materialisation shaped by it (Randerson, 2018).
Diagram A — Two-Node Spatial Diagram
Hot-Desert Node
Polar Node
Click any row above (Sound, Light, Tactile, Smell, Spatial) to expand it here.
Thesis. Draws on Bennett's (2010) vibrant matter and the sensory-turn literature (Howes, 2003; Pink, 2009; Ingold, 2000) to argue that a spatial, embodied threshold-crossing — walking bodily from a hot zone into a cold one — can materialise the project's comparative logic (RQ3) in the visitor's own body, not only in curatorial text. The central plinth operationalises the proposal's paired-object device (scent jar beside satellite tile) as a physical anchor between the two zones. People do not understand environmental change only through language, numbers and vision — knowledge also develops through bodily movement, temperature, texture, sound, smell, proximity, and emotional response. The project does not claim visitors are literally feeling the satellite data; they are encountering a sensory materialisation shaped by it, and no single channel here is asked to carry the whole meaning on its own.
Feasibility. Buildable from materials already specified in the field-operations workbook: directional speakers or bone-conduction panels per zone; standard dimmable gallery lighting on separate circuits; a radiant floor mat or heat lamp for the desert zone and a cooled floor panel or directed air vent for the polar zone; the seven-method captured objects (wax tablets, jars, charcoal cloth) displayed at fixed stations. A first prototype needs no custom electronics — thermal and lighting effects can run on simple timers rather than live data.
Open question to test. Whether a genuinely felt (not just intellectually understood) temperature or airflow difference is achievable within standard gallery HVAC and fire-safety limits. Worth testing at small scale — a single mocked-up threshold, not the full room — before committing to a final spatial design.
Diagram B — EO Data to Multisensory Translation
Reads across five EO datasets (four from the proposal, plus aerosol optical depth as a desert-specific addition) against five sensory-spatial modalities the installation may work in. Each mapping below is a speculative starting point, not a fixed design — one dataset, reasoned through five possible sensory logics at once.
Thesis. Sits within Gabrys' (2016) argument that EO sensing is never a neutral window but an active, constitutive practice — the diagram's "translation apparatus" does not claim to reproduce the environment exactly or reveal a sound, smell or texture hidden inside the data. It makes visible the chain (satellite measurement → value → artistic rule → sensory form) through which a measurement becomes a sensory form, treating data as material for situated interpretation rather than a message simply decoded. This directly targets the gap named in the Background and Rationale: existing climate-data art is dominated by visual and sonic modes and has barely engaged smell or touch (Randerson, 2018) — and where it does reach for smell, that channel needs the most explicit artistic framing of all five, since no satellite measures scent.
Feasibility. Each channel maps to a technique already prototypable at low cost: sound (pitch-mapping in free, open-source software such as Pure Data or Max/MSP, already standard in sound art); light (DMX-controlled LED fixtures, standard gallery kit); tactile (the wax-tablet cast objects themselves, static rather than actuated for a first version); smell (the enfleurage and charcoal-cloth captures, released by passive diffusion or a simple fan); spatial (a single off-the-shelf projector onto the floor). All five can be built and tested individually in the studio before any attempt at integration.
Open question to test. How much of this needs to be data-driven in real time, versus captured-and-fixed, to still read to an audience as meaningfully connected to the dataset. An early studio comparison — a static versus a data-driven version of the sound channel alone — would help settle this before scaling to all five channels.
Diagram C — Speculative Near-Live Earth-Observation Apparatus
A further-horizon schematic, offered for critical discussion rather than as a committed build. Called "near-live," not "live" — the system would not download data at the instant a satellite passes overhead (processed products are rarely available that quickly); it periodically checks a data service and updates when a new processed observation becomes available.
Sets the speculative translation work in Diagrams A and B within the project's actual research pipeline: both nodes run the same seven-method field protocol in parallel, feed a shared studio translation process, and converge — via the paired jar-and-satellite-tile object already specified in the proposal — into the storytelling labs and comparative analysis addressing RQ2 and RQ3.
| Stage | Hot-desert node | Polar node |
|---|
Comparative research questions (Storytelling Labs)
What the two-site comparative analysis stage is actually asking:
- Do desert and polar environments produce different metaphors?
- Do visitors describe environmental change as loss, danger, resilience or transformation?
- Does a near-live system make the issue feel more immediate?
- Do people trust something more because it is described as near-live?
- Does sensory experience change how they understand satellite information?
Thesis. The project's most speculative diagram, offered in the spirit of Randerson's (2018) "systems view" and Jasanoff's (2010) sociotechnical imaginaries: a near-live-updating installation would materialise the idea that satellite data is a continuously produced, present-tense infrastructure rather than a historical record, testing whether that quality itself changes the planetary imaginaries audiences articulate (RQ2). See "A systems view" above for the fuller network this diagram is trying to make visible, not just a speaker-and-light interface.
Feasibility. Technically achievable using the free, documented APIs already listed in the workbook's EO_Data_Reference sheet (NASA Earthdata, Copernicus Climate Data Store). The controller periodically checks the data service and updates the installation when a new processed observation becomes available — not at the moment a satellite passes overhead. A low-cost single-board computer (e.g. Raspberry Pi) drives standard actuators — a MIDI- or OSC-controlled speaker, a DMX dimmer, a solenoid scent valve of the kind used in scent-marketing installations, and a projector. Tactile elements remain physically fixed in this first prototype rather than electronically actuated — the same wax-tablet objects from Diagrams A and B, not a fifth near-live-driven device. The scent valve also controls timing and intensity only: the scent material itself is a site-associated composition stored separately in the valve, never something derived from the satellite reading (see the interpretation note at the top of this page). No custom hardware fabrication is required; the real cost is integration, calibration and reliability testing — see "Is it feasible?" above for the fuller list.
Open question to test. Real satellite revisit cadence (often one to sixteen days depending on the dataset) may in practice update slowly enough that a near-live system reads as static within a single gallery visit — see "The main problem" above for a proposed Fixed-vs-near-live test design. Worth testing directly against the Table 2 paired-object model, which does not depend on real-time updating, before committing resources to a near-live build.
Diagram D — Speculative Physical Installation: Tracing Particle Origin to Audience
A newer, more speculative extension of the same chain used throughout this page — added here as a pre-emptive record of the idea, not yet a committed design. Diagrams A–C translate a satellite reading, taken at the node, into a sensory form. This one asks a further question: rather than only showing what the air currently measures, could the piece also show where that air's particulate load actually travelled from before reaching the screen — turning the desert and polar particle fields into a literal, modelled origin story rather than only a live reading?
Two real, already-documented journeys — one per node
This isn't a hypothetical mechanism. Both nodes already have a real, named, scientifically tracked long-range transport phenomenon that fits this exactly:
Desert node — Saharan dust transport
Mineral dust lifted from the Sahara is continuously tracked by Copernicus CAMS and NASA (MODIS/CALIPSO) crossing the Atlantic toward the Caribbean and the Americas, particularly through the northern-hemisphere summer — a real, publicly documented transatlantic journey, not an artistic invention.
Polar node — Arctic haze
Industrial pollution from mid-latitude sources (chiefly Russia in winter, Europe in spring) travels north along isentropic surfaces and becomes trapped by strong temperature inversions — a well-established phenomenon in atmospheric science, meaning aerosol measured at a remote polar site can be traced back to specific industrial regions thousands of kilometres away.
Is it feasible?
The relevant technique — back-trajectory analysis, computing where an air mass came from over the preceding hours or days — is not experimental; it's the standard method atmospheric scientists already use for pollution source attribution, most commonly via NOAA's HYSPLIT model. What makes it a genuine open question for this installation, not a solved problem, is everything around running it continuously and honestly in public:
- HYSPLIT is not a live API — a fresh trajectory has to be computed, not fetched instantly
- the underlying meteorological data (GDAS) is openly archived but has to be pulled and processed
- a computed trajectory is a modelled estimate of air-mass movement, not a literal trace of specific particles
- captions must say "modelled back-trajectory," never imply certainty the installation can't support
- a lower-risk first version can replay real, already-published transport case studies rather than computing new ones on demand
See the installation guide (04_Installation_Design/Spatial_Immersion/Dual_Screen_AirQuality_Installation_Guide.md) for the fuller build plan, and the accompanying grant proposal for how this would be resourced.
Thesis. Extends Randerson's (2018) systems view one stage further upstream: rather than starting the chain at "processed dataset," this version starts at the emission source itself, making the invisible geography of air pollution — who is downwind of whom, and by how much distance — part of what the piece communicates. It also sharpens Gabrys' (2016) point that environmental sensing is a constitutive, infrastructural practice: a back-trajectory is not the air's history revealing itself, it's a model's estimate, built and presented as such.
Feasibility. The desert and polar nodes already have real, citable long-range transport phenomena to draw on (Saharan dust; Arctic haze), so the narrative content doesn't need to be invented — only sourced and captioned honestly. A first version can cite real, already-published transport events rather than compute anything live; a further version could run periodic (not literally live) HYSPLIT back-trajectories ending at the installation's location, in keeping with this project's existing "near-live, not live" discipline.
Open question to test. Whether naming a specific, real source region measurably changes how audiences respond compared with an unattributed reading — the same kind of fixed-vs-near-live comparison already proposed for Diagram C, applied here to attributed-origin vs unattributed-reading instead.