What Is the Difference Between a Sensor Reading and a Forecast?
From the earlier reference collection. Read the current software blog.
An observation reports a measurement made at a past time; a forecast estimates a future condition using a model.
An observation describes a measurement
A sensor reading is an observation made by an instrument at a particular place and time. It may be transmitted or displayed later, but its observation timestamp identifies when the measurement was taken. A reading therefore answers a limited question: what did this sensor report at its location at that time? It does not automatically describe the whole surrounding area.
For instance, a probe may report 16.8 °C at a stated depth at 09:00. If the dashboard receives it at 09:07, it remains a measurement from 09:00. Clear labels preserve that distinction and help readers avoid mistaking a delayed observation for a live condition.
A forecast estimates what may come
A forecast is an estimate of a future condition produced by a model or forecasting process. It depends on inputs and assumptions, and it refers to a target time or period rather than a completed sensor observation. A forecast can be useful for planning, but its status and horizon should be visible.
Imagine a system showing a measured value at 09:00 and a model estimate for 12:00. Those values may appear together on a chart, but the chart should distinguish their origins. Calling both “temperature” without marking which is observed and which is forecast invites readers to treat an estimate as a measurement.
Label source, time, and method
Use distinct labels such as “observed” and “forecast,” then show the relevant timestamps. For observations, give the observation time and, where useful, the sensor or station. For forecasts, show the forecast issue time, the time being forecast, and the model or source when that information is available. If a value is an estimate of the current state rather than a future prediction, label that separately as an estimate.
A legend should explain lines and symbols in plain language. For example, solid dots might represent reported sensor samples, while a dashed line represents predicted values. Do not rely on color alone. If a forecast is revised, retain enough provenance to tell which forecast cycle a past display represented.
Keep comparisons meaningful
Observed and forecast values can be compared to study model performance, but they must be aligned by location, time, and units. Compare a forecast for 12:00 with an observation made around 12:00, not with the most recent observation from several hours earlier. Record missing observations too; otherwise, a chart may make model evaluation look more complete than it is.
A useful dashboard gives each value a clear status and leaves uncertainty visible. Sensor data may be delayed or unavailable; forecasts may change as inputs change. Keeping those categories distinct helps a reader choose the right evidence for the task, whether checking what was measured or planning for what might happen.