Digital twin with real sensors: the case that connects the model to live data

We have explained here what an Environmental Digital Twin is: a living model fed by sampling campaigns, where every measurement arrives with its date, method and responsible person. Until the next campaign shows up, though, the model has no news from the world — alive in memory, but silent. The most faithful portrait of the site is still the one in the latest report.

This case changes that routine in a specific, demonstrable way: the twin starts receiving real sensor data on its own. It is not a promise for the future, nor a race dashboard with readings every second. It is a case running end to end in the app: live weather from two Brazilian cities flowing into the model without anyone typing anything, plus a ready slot for the customer's own telemetry.

The difference between a photograph and a window

A twin fed by campaigns is like a portrait taken on each technical visit: faithful at the instant it is taken, silent between one visit and the next. With sensors connected, the portrait becomes a window: you can look at any moment and see what is happening now — or as close to now as the data source allows.

What changes in practice is not the drawing of the site, but how often news arrives from it. And the usual honesty applies: this window opens every few minutes, not every second. Anyone promising second-by-second readings in environmental monitoring is usually selling the photograph of a dashboard, not a real operation.

What the case shows

The Digital Twin with Real Sensors case sits in the app's Demos panel. The scene brings an area of interest with one well and three stations:

1. São Paulo station. Live weather for the city: temperature, humidity, pressure, wind speed and rain.
2. Manaus station. The same parameters, from the other side of the country. At the same moment, the two cities can be living opposite weather — and the twin reflects each one its own way.
3. Multiparameter station. pH, conductivity and water level — parameters that a customer's field station measures for real. In this case, it appears on purpose with no reading, for the reason explained next.

The first two stations fetch data from the public Open-Meteo network, a free weather data service that requires no key and no sign-up. It is real data from an actual meteorological network — not a number invented for the demo. The trade-off is clear: it is the network's measurement for each city, not a reading taken inside your site by your own equipment.

The station with no reading — and why that is on purpose

The water parameters (pH, conductivity, level) only exist when there is a connector: the address of the customer's own telemetry service, from which the multiparameter station's readings arrive. Without a connector, there is no reading — and the case shows that gap instead of hiding it.

That is deliberate. The worst defect of a demo is pretending the data exists. If the model displayed some made-up pH there, the visitor would learn to trust numbers with no origin. By showing the gap, it teaches the opposite: in ecbyts, every number has a source — and a missing source shows up as missing.

How often the twin refreshes

Refreshing is automatic and configurable between 5 and 30 minutes, and the app polls with restraint instead of hammering the data service. The weather network itself renews its measurements roughly every 15 minutes — asking for more than that brings no new information, only traffic.

In other words: near-real time, in the honest sense of the phrase. Enough to watch the afternoon change in both cities, not to gauge a gust of wind.

A reading becomes a dated observation

Every reading that arrives from a sensor does not sit loose on a display. It enters the model as an observation: value, unit, date and time, source (which station, which network) and the trail of who registered it. It is the same discipline the app already applies to sampling campaigns — now extended to data that arrives on its own, several times a day.

With that, history piles up quietly: tomorrow, the model still knows what the station measured today, with date and source to check. The twin stays fresh without losing its past.

When the model waves

With readings arriving in a series, the app follows the trend. When a sequence drifts away from the expected behavior — climbing or falling consistently — it lights the alert on the surface where the trend appears. An alert here is an invitation to look, not a verdict: the person still interprets the reading, with the context of the site that only someone who follows the ground has.

What it is good for in practice

• Following the site's weather in near-real time, without waiting for the next field campaign;
• Keeping the twin fresh between campaigns, with the scene reflecting the present instead of the latest report;
• Recording readings as dated observations, with source and responsible — the same registration trail as the campaigns;
• Being warned when a trend drifts, instead of finding out weeks later in a spreadsheet.

What this case is not

• it is not per-second real time: the window works in minutes;
• the weather comes from a public network (Open-Meteo): real data for the city, not a measurement from your equipment inside your site;
• the water parameters depend on a customer connector: without one, the model shows the gap, it does not invent a number;
• the scene is a demonstration: the data is live, the setting is didactic;
• none of this replaces field sampling, laboratory reports or the interpretation of the responsible professional.

---

See it in practice

The guided tour walks through the case step by step, in about 10 stages, one action at a time, in language for someone who has never opened the app. It shows where the stations live, how automatic refresh is switched on, how a reading becomes an observation and where the trend alert appears. It runs in the browser, over the demo case, with no account needed.

Open the Digital Twin with Real Sensors interactive tour

For those who look after a real site: which reading would you like to see arriving on its own in your model — and how often?