~ $ cd projects/sea && cat README.md

The sea, a day at a time

Every day, NOAA puts together the temperature of the surface of every sea from what satellites, ships and buoys measure, on a grid of quarter degrees: its daily Optimum Interpolation analysis, OISST. This site keeps four of its cells, off L'Estartit, Barcelona, Tarragona and the Ebre delta, a value a day since 1982, and the year still running up to three days ago.

What follows is a blueprint: the pictures are drawn by nodes wired together, and the dials on the board turn them. Its menu holds a few more questions — this year day by day, the four points side by side, and whether a warm sea goes with warm nights.

off = dial "Off" value: barcelona
warmth = dial "Warm from, °C" value: 26
years = sea-years point: off warm: warmth
whole = keep "Only the whole years" table: years column: whole is: equals value: yes
lines "The sea's mean, year by year" table: whole x: year y: sst
trend = trend table: whole y: sst
readout "Each decade" value: trend.per-ten unit: °C about: "the change along the straight line fitted through the whole years"
bars "Days at least as warm as the dial" table: years x: year y: warm faded: whole
warmer = trend table: whole y: warm
readout "Warm days, each decade" value: warmer.per-ten unit: days about: "the change in the days a year at least as warm as the dial"

off = dial "Off" value: barcelona
days = sea-days point: off
lines "This year, and the normal for each day" table: days x: day y: sst and: normal
above = summary table: days column: above
readout "Above the normal, on average" value: above.mean unit: °C about: "the mean of each day less its normal, this year so far"

months = sea-months point: all
names = sea-points
named = join left: months right: names
whole = keep table: named column: whole is: equals value: yes
summers = keep table: whole column: season is: equals value: summer
years = group "A summer a row" table: summers by: year and: name value: sst how: mean
lines "Summer's mean, a line a point" table: years x: year y: sst split: name

sea = sea-months point: barcelona
nights = weather-months station: WU
both = join left: sea right: nights
whole = keep "Months measured whole" table: both column: whole is: equals value: yes
seasonless = season "Less each month's mean" table: whole by: month
odd = season "Less each year's mean too" table: seasonless by: year
measured = correlation table: whole x: sst y: tn
left = correlation table: odd x: sst y: tn
readout "As measured" value: measured.r about: "r of the sea's monthly mean and the mean night, month by month"
readout "The season and the years taken out" value: left.r about: "r of what is left of each"
scatter "What is left of each month: the sea against the nights" table: odd x: sst y: tn colour: season

What is in it

• The sea is warmer at all four points than when the record starts. Off Barcelona, the mean of a year's days was 17.6 °C over 1982–1991 and 19.1 °C over 2016–2025. The straight line through the whole years rises between 0.38 and 0.45 °C a decade at the four points, and the summers' between 0.50 and 0.63.
• Days at 26 °C or more have gone from rare to usual, except in the north. Off Barcelona there were 1.3 a year over 1982–1991 and 34.7 over 2016–2025, and 2025 had 68; off the Ebre delta, 14.9 became 50.2, and 2025 had 87. Off L'Estartit, the coolest of the four, the sea seldom gets there: four days in 2006, its most.
• The warmest year was 2025 off Barcelona, and 2022 off the other three. The warmest day off Barcelona was 23 August 2023, at 28.73 °C; off Tarragona and the Ebre delta, 10 August 2003.
• A warm sea goes with warm nights. Off Barcelona, against the nights at Badalona, month by month, with the season and the years taken out, r is 0.45, and 0.76 in summers alone: a month whose sea was warmer than usual for its season and its year had warmer nights than usual too. The other pairs of a point and a station by the sea point the same way.

What it is not

A cell is about 25 km across, and is the open sea off a place, not the water at its beach: the one off Barcelona has its centre some fifteen kilometres out. OISST is an analysis, not a thermometer — where there is no measurement, the grid is filled in from what is around it — and the dataset itself warns that values less than fifteen days old may still be revised. A finished year is kept only from the middle of the following January, once they are not.

Every figure above is of the whole years, 1982 to 2025. The year still running is drawn day by day in the blueprint's second question, beside its normal, and is kept out of every one of them: it is not over, and its last fifteen days can still change. A trend is the straight line through the whole years, and says what the record did, not what it will do, nor why; a correlation says what goes with what, never why.

Where it comes from

NOAA's daily Optimum Interpolation Sea Surface Temperature, version 2.1 (Huang and others, 2021), from the server of NOAA's Physical Sciences Laboratory: data provided by the NOAA PSL, Boulder, Colorado, USA, from their website at psl.noaa.gov. The site keeps the days of four cells, a finished year at a time, and asks for the year still running again every day; the weather beside it is the Meteocat's, as on the page of hot nights.