Why remote stations lose data, and how to design around it
Sensors rarely fail first. Power, communications and fouling do. A station designed for those three keeps its record.
After a monitoring network has run for a season, the gaps in its record usually trace back to the same few causes. None of them is the sensor.
Power
A solar panel sized for a clear week in March will not carry a station through two overcast weeks in July. Size the battery for the longest cloudy spell you expect, not the average day, and log the battery voltage as a channel in its own right so that a failing battery is seen before it stops the station.
Communications
Mobile coverage at a riverbank or on a char is weaker than a coverage map suggests, and it changes with the season. The logger must store readings locally and forward them when it reconnects. A station that only transmits live readings loses everything measured while the link was down.
Fouling
In warm, sediment-laden water, optical and electrochemical sensors foul within weeks. Wipers and copper guards help. A scheduled cleaning visit, with a check reading against a reference, helps more.
Lightning and water
Surge protection on every cable that leaves the enclosure, proper earthing, and connectors rated for immersion where a flood can reach them. An enclosure mounted above the highest recorded flood level is the cheapest insurance in the project.
Knowing that something is wrong
The portal should show when each station last reported and raise an alert when one goes quiet. The worst gap is the one nobody notices for a month.
What this means for a specification
Ask for local storage, a stated power autonomy in days, monitoring of the station's own health, and a maintenance schedule with named responsibilities. These lines protect the record more than any sensor accuracy figure.