Free Metocean Data for Weather Window Planning
Every offshore proposal has a line for weather downtime, and every survey plan has to answer the same question: which month should we go? Too often the answer is a guess, or a number copied from the last job in a different sea. Yet good metocean data for planning is free. It just takes some work to download and turn into something useful.
I built a small Python tool that does that work for one project location. You enter a coordinate and your operational limits, run it, and get monthly wave and wind statistics, current statistics, wave and wind roses, a short-term wave forecast, and a weather window analysis, all in one Excel file plus charts. This article explains what data it uses, what the weather window numbers mean, and where the limits are.
Where the data comes from
All the data comes from the Copernicus programme, through the WEkEO platform, and is free to use with a registered account:
- Waves and wind: ERA5, the ECMWF reanalysis. Hourly significant wave height (Hs), mean wave period, wave direction and 10 m wind, on a grid of about 28 km, going back decades and updated to within about a week of today.
- Currents and sea temperature: the Copernicus Marine global physics model. Daily currents and temperature at each depth, on a grid of about 9 km, covering the last few years and a few days ahead.
- Wave forecast: the Copernicus Marine global wave model. Hs, peak period and direction every three hours, about nine days ahead.
ERA5 is a reanalysis, not raw measurements: a weather model run over the past and constantly corrected with observations from satellites, buoys and ships. That makes it consistent over long periods, which is exactly what you need for monthly statistics.
How the tool works
The settings live in one plain text file, so nobody has to edit code:
- Project name, which becomes the output folder name.
- Latitude and longitude of the site in decimal degrees (WGS84).
- Operational limits: maximum Hs, maximum wind speed and the minimum length of a working window.
- Which datasets to download, and for which period.
Then you double-click a batch file. The tool downloads the data for the nearest grid point, computes the statistics and writes everything to a results folder. Data that was already downloaded with the same settings is not downloaded again, so changing only the limits and running again takes seconds.
Two details turned out to matter:
- If the nearest grid point is on land, which happens near coasts and islands, the tool moves to the nearest sea point and reports which point it used. Silently using a land cell would give zero waves and nonsense statistics.
- If something fails, such as a timeout from the server or a login problem, the full error is written to a text file in the project folder, while the screen shows a short message.
What the operational limits mean
The weather window analysis uses three numbers. They are the most important input, and also the easiest to get wrong.
Maximum Hs. Significant wave height is the average height of the highest third of the waves, roughly what an experienced observer would call the wave height. Typical limits are:
- Multibeam survey from a small vessel: about 1.0 to 1.5 m.
- ROV or diving from a survey vessel: about 1.5 to 2.0 m.
- Larger vessels, for example cable or pipe lay: up to 2.5 to 3 m.
The right number for a project comes from the vessel's operational limits or the operations procedure, not from a table like this one.
Maximum wind. Wind speed at 10 m above the sea, in knots. Twenty knots is about 37 km/h, a fresh breeze with white caps.
Window duration. How long the safe conditions must last without interruption for the job to be done. One survey line, one ROV dive or one lift might need 12 hours. If the sea is calm for 8 hours and then gets rough, that calm period does not help, even though every one of those 8 hours was safe.
Two different percentages
The tool reports two numbers for each month, and the difference between them is the main point of the analysis.
Safe time is the percentage of all hours in the month when Hs and wind are both within the limits.
Window probability is the percentage of hours when you could start work and still have at least the required number of safe hours in a row ahead of you.
The second number is always equal to or lower than the first. In a test at a point in the Java Sea, with limits of 1.5 m, 20 knots and a 12-hour window:
- In April, both numbers were 100%: the sea never exceeded the limits.
- In December, 94% of all hours were safe, but the chance of being able to start a 12-hour job at any given time dropped to 90%.
The gap between the two numbers grows with longer windows and rougher seasons. For a job that needs 24 or 48 hours without interruption, it can become large, and that is the number that drives real downtime.
What comes out
For each project, the results folder contains:
- One Excel file with every table on its own sheet.
- Monthly wave and wind statistics: mean, median (P50), 90th percentile (P90) and maximum Hs, mean wave period, wind statistics, safe time and window probability.
- Monthly charts of Hs and wind with the operational limit drawn as a line.
- A weather window chart per month.
- Wave and wind roses (direction the waves and wind come from).
- Daily current and temperature statistics for the surface and for the water column average, plus a current rose (direction the current flows to).
- The short-term wave forecast with working windows marked in green, and a list of each window with start, end and duration.
All times are in UTC, which is worth stating clearly in any report, because local time is easy to assume.
How I use the numbers
- Choosing the season. Comparing safe time and window probability month by month shows which months are worth bidding for and which need a weather allowance.
- Estimating downtime in proposals. One minus the window probability for the planned month and job duration is a reasonable first estimate of weather standby, and it is defensible because the data source is named.
- Checking a plan before mobilisation. The nine-day forecast with marked windows is a quick way to see whether the coming week has any usable windows at all.
- Talking to clients. A chart with the limit drawn on it is easier to discuss than a paragraph about "the monsoon season".
Limitations
The data is good, but it is not everything:
- Global models are least reliable near coasts, in bays and in narrow straits, where local effects the grid cannot see dominate.
- ERA5 is smoothed in time and space and tends to underestimate short, sharp peaks such as squalls.
- The results are for planning, proposals and weather window estimates. They do not replace a design metocean study or measurements at the site.
- The data source must be credited in any report, as required by the Copernicus licence.
What I took from building it
- The limits are the real input. The same data gives completely different answers for 1.0 m and 2.0 m, so the limits should come from the vessel and the procedure, not from habit.
- Duration matters as much as wave height. Safe time alone is optimistic; the window probability is closer to what the vessel will experience.
- Make it easy to run. A text config file and a double-click are the difference between a tool that is used on every proposal and one that is used once.
- Say what the numbers are not. Being clear about model limits makes the rest of the analysis more credible.