A squint at Met Office HadNMAT2

Posted: June 25, 2015 by tchannon in Analysis

Given 28 gridded datasets and a variety of novel things to try, I be flummoxed. Try this

The new dataset kid on the block, seems to have appeared 6th June is Met Office Hadley Centre HadNMAT V2, nighttime marine air temperature. Grided at 5 degrees.


Figure 1, an experimental plot from data generated by alpha software. So far as I know this is correct. The main graphic is a hovmoller plot intending to show how the contribution to a global temperature computation has varied spatial with time.

Looking at that lets eyeball a month where there seems more southern data, typed in 1890, left the month at 4 from something else, random,


Figure 2, the active gridcells. We can see the pattern of shipping traffic used for the dataset. Both the Suez canal and both Capes in use, not Panama Canal. Trade and military activity in Indochina shows.

The pattern will change as technology, trade and warring tribes take their toll.

Out of idle curiosity, see that red cell off Newfoundland? The Atlantic is well covered so lets extract the time series, one better all 9 cells as a block with 47.5N 47.5W at the centre. Easy to work out cell centres via the PDF plot, or can plotted with annotation.


Figure 3, Set of 9 gridcells of 5×5 degrees.



Figure 4, plot of gridcell time series in one.

Ships rarely sail in a circle for a month yet these are monthly data points. Sharp temperature differences between adjacent cells are improbable, the scatter is huge.


Figure 5, Standard Deviation for the 9 cells against time. WWI and WWII are obvious. Consistent it is not.

Sorry about being boring, obsessive, takes a lot of persistence to the exclusion of all else to do significant software.

Post by Tim

  1. michael hart says:

    Via the wiki-thing I read:
    “In 1880, the Oregon Railroad and Navigation Company steamer, Columbia, became the first application for Edison’s incandescent electric lamps (it was also the first ship to execute use of a dynamo).”

    Hmm… So, compared to later years, in 1875 did sailors find it more difficult to read thermometers accurately at night than during the day? Or, put another way, does the night-time temp series have the same or larger error bars than the day-time series?

  2. tchannon says:

    A good question is why night temperatures when plainly day temperatures exist.

    The whole thing is a bit of a joke, so much so that part of the dataset is a height time series which I have not extracted. The size of ships has changed, the speed has changed, the routes have changed, the power source has changed, everything has.

  3. Dear Michael, tchannon,

    Day time temperatures are not used because when the sun’s up it heats the ship and artificially raises the air temperature. During the night, the ship cools so the night temperatures are a more accurate estimate of the true air temperature.

    There’s a height time series because temperatures drop as the you move vertically up from the ocean surface. As ships have got higher with time, they measure at a greater height above the surface. It’s not a massive effect – a few tenths of a degree – but it’s one that can be estimated.

    Best regards,

  4. Paul Vaughan says:

    Tim, top-panel legendary z-axis scale-definition for HadSST3 & HadNMAT2 might improve potential for audience insight.

    Tim suggested:
    “Sharp temperature differences between adjacent cells are improbable, the scatter is huge.”

    I can often feel the temperature difference without even putting my hand in the water when I sea-kayak across an eddy-line.

    Caution: blending latitudes longitudinally leaves a key diagnostic untapped:

    “The uncertainty in the height adjustment due to uncertainty in the stability scales with the difference in height from the 10m reference height and is relatively larger in regions where the variability in the stability is larger (e.g., regions with western boundary currents, not shown). The contribution of uncertainty in the measurement height scales with the uncertainty in the height, but is greater for measurement heights low in the boundary layer where vertical gradients of temperature are larger.”

    Global analysis of night marine air temperature and its uncertainty since 1880:
    The HadNMAT2 data set


    US NOAA & UK Met Office Administrators: Failure (whether via ignorance or deception) to uphold basic standards of ethics & integrity probably isn’t the legacy you’re going for. Sober diagnostic reflection is due.

  5. Paul Vaughan says:

    John Kennedy: I’ve left you a friendly tip on the NOAA ERSSTv4 thread.