World map of regional temperature trends over decade periods using ERSST V5 and GHCN V4

This page shows a shaded WebGL plot of regional trends of monthly average temperature anomaly over multi-decade periods (you can choose). It uses unadjusted GHCN V4 with ERSST V5. It replaces a GHCN V3 version, in which the station mesh was shown, and station information was accessible. The old version is preserved here.

The new version is based on the Moyhu WebGL facility. It uses LOESS graphics; this has some important consequences, which I'll describe below. The main new feature in controls is that you can choose any interval with the buttons on the right. There is more discussion and operating advice below.


As usual, the sphere is a trackball that you drag into position, or you can quickly set by clicking on the small map at the top. Beside that map are checkboxes which let you switch the objects displayed. The icosahedral mesh and nodes are initially not shown.

The trend interval is chosen with the buttons on right, The colored buttons mark the range ends. At first, it shows 1980 and 2020; that means Jan 1980 to Dec 2019 (missing months are handled). The rules for changing are that if you click outside the range, the nearest button will move, but if you click inside, the red button will move. You can click the pink button to make it red. When you have chosen an interval, you must click the top "Show" button for it to take effect.

When you click on the globe, the trend at that point is shown under the checkboxes. The global average trend is shown beneath the buttons, but only after clicking "Show".

The map is created by first getting monthly averages on the 5762 icosahedral nodes, as described here. The trends are then calculated on those nodes. The LOESS method takes a weighted local linear regression on the closest station/SST data, even if it is not close at all. In Antarctica, for example, before 1960 that generally means ocean data. So trends for Antarctica for early times should be ignored. Elsewhere, there is loss of resolution according to station data, but it is still reasonably based. GHCN V4, of course, has much better coverage than V3.

Note that the color scheme is centered for zero trend, but the range varies with the length of period.


  1. Very nice Nick.

    I think it might be better if all trends were shown on the same colour scale i.e. trends over any period fixed at say -0.5 to 1.5C/decade irrespective of the actual trends.

    1. Thanks, Clive.
      I think there needs to be some flexibility of scale - trends starting 1907 have a much smaller range than starting in 1997. But it would certainly be a good idea to use the same scales for unadjusted and adjusted, for comparison. And I could arrange for the scales to change in just a few steps, going back in time. I'll do that.

  2. Nice visuals. People have also used Google Earth to make good animations e.g.

    1. Thanks, RB,
      I used to work woth Google Earth files - a recent post with links is here. But I foun dthat there was a lot of overhead, and not much flexibility. So I switched to uning Google Maps, eg here, where I can use Javascript. But I haven't tried animations.

  3. Nick, your imaging tools are superb and national meteorological and space agencies should be jealous.

    I am wondering if you have physical explanations for almost all the islands of the world showing significantly more warming than the surrounding sea when you look at adjusted data interval. The longer the interval the more pronounced the apparent bias. Also, although it makes sense that land should show more warming than sea surface, how do you explain 80-year-plus cold spots in every continent except Antarctica and Europe?

    Jones(2016) claims that a robust GMST trend can be obtained from as few as 100 globally equidistant stations or one station for 100 years. How does this jibe with these land cold spots and universal island warming?

    1. Thanks, Ron. Yes, we talked a little aboutr islands at Clive's. As you say, there is a pronounced discrepancy, even after adjustment. We know that overall land warms faster than sea; I can only think that this has a component that reflects the air phase, rather than just being over a continent.

      The spatial variations may reflect the way trend can be affected by a warm or cool spell near one of the ends of the period. But there could also sometimes be a component of some local measurement deviation.

      I've looked at the question of reduced measurement locations, usually under the heading of "just 60 stations". My latest was here, and it links back to earlier efforts. I think it is true that a much reduced but well sprad group of stations can give a good result. That recent post focussed on what happened if you forced the selection of islands rather than SST. It does make a difference.

  4. Thanks for your reply. Using the premise that there is a significant radiative imbalance over the oceans due to enhanced CO2 forcing being cooled by fresh (200-1000-yr-old) sea overturning, there should be significant divergence of surface temperature trend between the sea and the continents. After 50 to 100 years most all natural variability should have canceled out leaving a significantly defined centennial trends like long-term sea current oscillations (and AGW footprint). This being the case, on your graphic all seas should be solid green and the land solid orange. The continental coastal areas should be yellowish in the direction of the prevailing wind due to surface advection (wind driven heat transfer). This coastal effect is known as the marine effect and the reason west coast of the USA has a milder climate than the east.

    Under this physics all the non-continental islands should be the same color as the sea surrounding year-round, let alone a hundred-year interval. Nick, am I correct that you are postulating that the marine effect on islands is negligible to their climate? I believe the marine effect goes for hundreds of kilometers. Therefore almost every non-continental island must be considered a marine climate and the weather stations there should be reflecting the ocean anomalies in their own anomalies.

    Even if the weather station is 50 kilometers from the coast that the wind is coming from the land effect delta will have changed only a tiny percent compared to pre-industrial. If it is more then that then one would expect the centers of large continents that have the wind brought to zero radiative imbalance should show consistently steep 50-100-yr anomalies. But we find this is not the case from your graphic tool. In fact most of the continents have a warming hole. Nick, I hear you saying that the warming holes are due to the starting decade having an unusually high temperature in that region. However, there seems to be much more rapid fluctuation of variability then for a random ocean or land region to maintain and significant natural variation for 100 years, though I think this should be investigated.

    1. Ron,
      First I should emphasise that I'm just showing the results as they are. I have no particular opinion about how they ought to be. I can see why you would expect islands to follow the local sea temps, but you can only find out by checking. I actually don't think the discrepancy is as great as you suggest, but there is certainly some.

      Here is a plot of annual averages of Hilo, HI (black) and the nearest sea point. The trend 1907-2017 of HILO was 1.164 C/Cen; the sea was 0.609. They sometimes track well, sometimes not. FWIW, adjustment increased the discrepancy.

    2. Nick, I certainly appreciate that your only consideration with the data is its presentation in enhanced, view-able form, for which you have notably succeeded. As you know, I am skeptical of many of the consensus claims. OTOH, I try not to invest in a position, and thus am always curious when new tools or clues can be found to shed light toward either validation and falsification. The land record being warmer that the SST record, as you know, has been presented as a confirmation of AGW, showing evidence of the radiative imbalance caused by the ocean's 90% greater heat capacity as opposed to land's. I have never seen this assumption tested in studying for a direct relationship to the warming trend of temperature observation points relative to degree of land versus ocean character/proximity.

      Thanks for the Hilo plot comparison to nearest sea point. Understandably, there is a variation due to local precipitation and other effects, but over a hundred-plus years there should be discernible trends. And there seems to be, but they are not what I think one would expect from AGW theory. The adjusted land record is visually of higher trend throughout the ~120 years. And, the steepest rise by far in both plots is in the 1900-1960 interval rather than the 1960-2016 interval. Independently, there is a steady divergence of land from sea over the entire 120 years. Being a such a long record these surely are statistically robust trends but how do they fit in with AGW theory and the Keeling Curve?

      Hilo's National Weather Service station is on a peninsula that is only 1.5K from the sea at the shortest point and ~4K on average in 180 deg arc. This should make Hilo very much a marine climate. As for non-climate effects, looking with Google Earth I do not see any air conditioning condenser near the enclosure but I do notice that it's on an airstrip with lots of nearby black pavement and no trees until the surrounding community. I wonder if there are historical photos, ground or aerial view, of the station. There are about a dozen other stations in the Hawaiian Islands. Would you be interested in doing a post on comparing them to the nearby sea records? Personally, I don't understand why this hasn't been done routinely. You seem to agree there is something unexpected here. I am not certain of a discrepancy of data to theory just that I see there is an opportunity to check, a rare opportunity. Here is Clive Best's plot of Hilo anonmaly.