agr-r is an agricultural SaaS product built around a single defining artifact: a 3D digital twin of the entire Earth. The product's purpose is to let agricultural decision-makers observe the planet as a living, data-driven model and to see, in three dimensions, how water flows across it — runoff, accumulation, and discharge across terrain and watersheds.
The product intent, derived directly from the authoritative requirement thread, is threefold and non-negotiable:
Audience: Agricultural experts and decision-makers — agronomists, land managers, irrigation and watershed planners — who need to read a whole-planet digital twin and watch water move across it in order to ground agricultural planning and intervention decisions. The product is an instrument for serious observation, not a consumer toy.
agr-r delivers a browser-based, full-bleed 3D Earth twin as its primary surface. The globe is not a decorative render: it is a data sculpture whose arable regions glow, whose hydrology is rendered as luminous fields, and across which animated water-flow ribbons stream, pool in river basins, and discharge at watershed outlets. A thin HUD overlay — a top rail, a left instrument panel of layer toggles, and a right readout column — frames the canvas without competing with it.
Current delivery: A custom first-party web UI (custom_ui: true) with three pages — Landing, 3D Globe, and Water Flow — all reachable without an account (access_requirement: none). There is no application-owned identity, no sign-in, and no account management in the current scope. The product is observational and analytical: the user views the twin, toggles data layers, scrubs scenarios, and reads flow metrics.
Actors:
Accepted behavior: Viewing the whole-Earth 3D digital twin; observing 3D water flow across it; toggling data layers (hydrology, arable, elevation, precipitation); switching between named scenarios (baseline / drought / irrigation plan); reading flow metrics (discharge m³/s, runoff coefficient, basin area) that update on hover; and scrubbing the flow simulation along a hairline timeline rail.
Narrow exclusions: No stock photography of fields, tractors, or farmers. No flat 2D map fallback that abandons the digital-twin premise. No account, login, or role/permission system. No transactional, retail, or form-heavy workflows. No future-horizon features are included in current pages or acceptance.
agr-r is delivered entirely as a first-party web application. Every accepted capability — the landing presentation, the globe, and the water-flow observation surface — is owned by the application itself and is reachable anonymously. The user does not sign in, does not create an account, and does not resume private durable state; the product is a live instrument that presents the same planetary twin to every visitor. Consequently, no identity establishment, verification, invitation, or provisioning lifecycle exists in the current scope, and none is inferred, because no accepted journey requires durable actor-specific state, a bound commitment, or a value transfer to a specific participant.
The delivery boundary is deliberately narrow. The current product is the twin and the flow: a 3D Earth, its data layers, its animated water, its scenarios, and its readouts. Everything the user does happens on the canvas or in the HUD rails around it. There is no separate data-management, reporting, export, alerting, or collaboration surface in the current scope, and none is added by convention.
Future-horizon material, where it exists, is confined to Section 11 and is excluded from current pages and acceptance.
Each requirement below is a distinct story point with provenance, lifecycle facts, and observable acceptance. Provenance is explicit where the authoritative thread states it directly, and required_inference where the mechanic is indispensable to make an accepted outcome usable.
FR-1 — Agricultural-purpose application As an agricultural expert / decision-maker, I should use an application built for agricultural purposes, so that my observation work serves agricultural planning and intervention decisions.
FR-2 — Whole-Earth 3D digital twin As an agricultural expert / decision-maker, I should view a 3D model of the entire Earth as a digital twin, so that I can read the planet as a single coherent data artifact.
FR-3 — Visible 3D water flow As an agricultural expert / decision-maker, I should see the flow of water in 3D within the model, so that I can understand where water moves, accumulates, and discharges.
FR-4 — Layer toggles as instrument switches As an agricultural expert / decision-maker, I should toggle data layers — hydrology, arable, elevation, precipitation — so that I can re-weight what the twin shows me.
FR-5 — Scenario switching As an agricultural expert / decision-maker, I should switch between named scenarios — baseline, drought, irrigation plan — so that I can compare planetary water behavior under different conditions.
FR-6 — Flow metric readouts on hover As an agricultural expert / decision-maker, I should read discharge (m³/s), runoff coefficient, and basin area as I hover any point on the globe, so that I can quantify what I observe.
FR-7 — Timeline scrubbing of the flow simulation As an agricultural expert / decision-maker, I should scrub the flow simulation along a hairline timeline rail, so that I can move through the flow's behavior over time.
FR-8 — Landing presentation of the twin As an agricultural expert / decision-maker, I should arrive at a full-height generative hero that presents the Earth twin, so that my first impression is an observatory rather than a landing page.
FR-9 — Scroll-scrubbed camera and scenario morph on Landing As an agricultural expert / decision-maker, I should scroll the Landing page to move the camera and morph the flow field between scenarios, so that the page itself teaches me the twin's behavior.
Product context. This persona works with planetary-scale agricultural data and needs to read a whole-Earth digital twin as a single coherent artifact. Their work is observational and analytical: they watch water move across terrain, note where it accumulates and discharges, and translate those observations into agricultural planning or intervention decisions. They are not a consumer browsing a map; they are an expert using an instrument, and they expect the data sculpture to be the loud part of the screen while the chrome stays quiet enough for long sessions.
Primary goal. To observe the 3D water flow across the whole-Earth twin and use that observation to ground agricultural planning or intervention decisions.
Distinct accepted responsibilities.
Relevant inputs or decisions. The persona decides which layers to enable, which scenario to compare against, which points on the globe to interrogate for metrics, and where in the flow timeline to focus. Their inputs are the layer toggles, the scenario selector, the timeline scrubber, and the cursor position on the globe.
Interactions with other accepted participants. This is the sole accepted active human persona. There is no second human participant, no counterparty, and no recipient whose state changes as a result of this persona's work. The persona interacts only with the system's real-time simulation, layer re-weighting, and readout computation — non-persona system processes that support but never own the human-facing capability.
Observable success. An interpretable 3D water-flow display on the whole-Earth twin, with layer and scenario controls that visibly re-weight the flow field and readouts that quantify what the persona observes — sufficient to support an agricultural planning or intervention decision.
Source-backed constraints. The persona is inferred from the authoritative requirement thread, which asks for an agricultural-purpose digital-twin application in which the whole Earth is visible in 3D and water flow is visible in 3D. No biography, demographics, permissions, or account requirements are asserted, because the source does not establish them.
Failure/recovery: if the real-time canvas cannot initialize on Landing, the dark ground, headline, micro-label, and CTA remain readable and the persona can still proceed to the 3D Globe surface; the flow field is presented as a still frame with a manual play control.
Failure/recovery: if a layer fails to load, the globe remains usable with the remaining layers and the failed layer's toggle shows an inactive state. If a scenario fails to load, the previous scenario remains active and the selection is marked unavailable. If the canvas fails entirely, the persona is returned to the Landing surface with its still-frame fallback.
Failure/recovery: if the simulation cannot run, the flow field is presented as a still frame with a manual play control and the readouts remain available. If scrubbing is unavailable, the simulation continues from its current position and the rail shows a disabled state. If the canvas fails entirely, the persona is returned to the 3D Globe surface.

Resolving the Earth twin.
Scenario plates
Each plate names a scenario and its flow signature — the camera holds on the twin while the plates move past it below.
01/ 03active
Reference condition
The baseline holds present-day climate and land use, so every flow readout is a direct observation of the twin as it stands. Ribbons run dense across the temperate basins and discharge steadily at the watershed outlets, giving the comparison an unmodified reference for whatever follows.
Baseline discharge and runoff across the current hydrological year
02/ 03available
Water deficit projection
Drought re-weights precipitation and evaporative demand together, thinning the flow field and pulling accumulation back into the upper catchments. The outlets that carried the baseline volume now arrive late and low, so the gap between the two scenarios can be read as a single difference in the plate below.
Projected drought-year discharge against the baseline reference
03/ 03available
Intervention scenario
The irrigation plan adds abstraction and return flow on top of the drought signal, restoring discharge downstream without rebuilding the lost upland storage. The twin shows where the added water actually travels, which is the point of reading a plan on the planet rather than on a spreadsheet.
Modelled discharge under the irrigation plan intervention
Scroll to scrub the twin, or jump straight to a scenario.
Still frame
Holding reference condition on the frozen twin.

Resolving the Earth twin.
Scenario plates
Each plate names a scenario and its flow signature — the camera holds on the twin while the plates move past it below.
01/ 03active
Reference condition
The baseline holds present-day climate and land use, so every flow readout is a direct observation of the twin as it stands. Ribbons run dense across the temperate basins and discharge steadily at the watershed outlets, giving the comparison an unmodified reference for whatever follows.
Baseline discharge and runoff across the current hydrological year
02/ 03available
Water deficit projection
Drought re-weights precipitation and evaporative demand together, thinning the flow field and pulling accumulation back into the upper catchments. The outlets that carried the baseline volume now arrive late and low, so the gap between the two scenarios can be read as a single difference in the plate below.
Projected drought-year discharge against the baseline reference
03/ 03available
Intervention scenario
The irrigation plan adds abstraction and return flow on top of the drought signal, restoring discharge downstream without rebuilding the lost upland storage. The twin shows where the added water actually travels, which is the point of reading a plan on the planet rather than on a spreadsheet.
Modelled discharge under the irrigation plan intervention
Scroll to scrub the twin, or jump straight to a scenario.
Still frame
Holding reference condition on the frozen twin.
No comments yet. Be the first!