Observing a coastal port: a reference design

01What a port needs to see

A port observing site answers three questions at once: what the weather is doing, what the water is doing, and what is moving on it. Wind, visibility and precipitation decide whether cranes may work and pilots may board, and the methods for measuring them are set out in the international guide to meteorological instruments¹.

Water level and sea state decide under-keel clearance and berth safety. Sea level is measured with tide gauges, and radar gauges mounted above the water are now a common choice because they are simple to install and maintain². Long records of sea level are collected and shared internationally, which lets a port compare its own record with the region around it³.

Traffic completes the picture. Ships report their identity, position and course over the automatic identification system, a radio service defined in an international recommendation⁴, and vessel traffic services set out what a port must be able to see and do with it⁵.

Each of these is measured by a different instrument, at a different rate, by a different team. The value of an observing site is that it brings them onto one clock and one record, so that a crane supervisor, a pilot and a harbour master are looking at the same moment in the same way.

02Siting each instrument for what it must see

Where an instrument stands decides what it measures. Wind should be measured clear of buildings, cranes and stacked containers, which bend and slow the flow around them, so the guide to observing practice sets out exposure rules for each kind of measurement¹. On a busy quay that usually means a dedicated mast at the windward edge of the terminal rather than a sensor bolted to the nearest roof.

A radar tide gauge looks straight down at the water from a rigid mount. It must be clear of mooring lines, ship hulls and wave splash, and it must be tied to a surveyed height reference so that its readings can be compared over decades². Without that tie, a sea-level record cannot be joined to the regional records it should be read against³.

The buoy sits offshore where it meets the waves before the breakwater changes them, and the coastal radar is placed high enough to see the approach channel past the cranes. Each choice is recorded with the instrument, so anyone reading the data later knows where and how it was measured.

03One network for sensors, cranes and people

The instruments, the cranes and the people on the quay share one private mobile network on the port's own premises. Mobile network standards define how such non-public networks are set up and kept separate from public ones⁶.

Every measurement depends on a shared clock. The site keeps time with a precision time profile carried over the network, so radar plots, buoy reports and crane events line up to the same instant⁷.

Computing sits beside the quay rather than in a distant data centre. Decisions that cannot wait, such as stopping a crane in rising wind, are made on site, and the same values are forwarded upstream for planning and records.

The radio links, the clock and the edge rack are designed together, because each depends on the others. If the network carries the time, it must be built to carry it accurately; if decisions run at the edge, the edge must keep its own copy of everything it needs to decide.

04Values that keep their context

A reading on its own is a number. A reading with its time, place, instrument, method and quality flag is an observation that someone else can trust. Each value leaving the site carries that context, using an open interface for sensor data so that any system can read it⁸.

Weather and water observations can also be shared with national and international services. The global system for exchanging meteorological data is moving to open web standards and a publish-and-subscribe model, so a port can contribute without special links⁹.

Quality control happens at the edge, close to the instrument, and the checks applied travel with the value. When a sensor drifts or fails, its readings are flagged rather than silently dropped, and corrections are published as corrections.

05When the links fail

Storms that make the observations matter most are also the ones most likely to cut the links. The reference site is designed to keep observing, deciding and recording locally while it is cut off, and to hand on everything it stored once the link returns.

Local decisions keep running on the edge rack, the shared clock holds its time through the outage, and stored observations are replayed in order with their original timestamps, so the record has no gaps and no duplicates.

The same design also covers the opposite case: when an instrument fails but the network does not. Its absence is recorded as a gap with a reason, never filled in silently, and the operators see at once which picture is incomplete.