Tuesday, July 13, 2010

Arctic Trends using GSOD Temperature data with TempLS

The previous post described the GSOD land temperature data from NCDC. This is a large set based on SYNOP data, which is plentiful from about 1973 onwards. Ron Broberg has processed it and put it into a useful format.

In particular it has a lot more Arctic stations than does the GHCN set. So my first regional application with TempLS will be to Arctic trends.



I defined two regions - Arctic N of 60°, and HiArctic N of 70°. I ran TempLS using both GHCN and GSOD - GHCN coverage was sparse. The trends for GSOD were:

Trend °C/DecArGSODHiArGSOD
Number1935244
Trend 1979-20090.56940.9063
Trend_se0.086640.1272

and for GHCN:
Trend °C/DecArGHCNHiArGHCN
Number29447
Trend 1979-20090.46770.6971
Trend_se0.088940.1016
Here are the side-by-side plots. The scales do not necessarily match, and the temperature offsets are not the same for GHCN and GSOD.

Arctic GSODArctic GHCN
High Arctic GSODHigh Arctic GHCN


Both datasets show strong warming since 1979, especially in the High Arctic. The more numerous GSOD data shows a greater trend (0.9 °C/Decade).

3 comments:

  1. Still noisy, though.

    The jump in the high arctic around 2005 stands out a bit. Might be worth doing an EM Smith test there - make sure the stations that dropped off around that time weren't trending differently than the ones that stayed.

    ReplyDelete
  2. Well, even tho GHCN has few high arctic stations, it and GSOD show the same jump, and I don't see a dropoff in GSOD stations at that time.

    I'm planning to do a regional trend plot - but I need a new set of orthogonal functions for it.

    ReplyDelete
  3. When Arctic loses ice, winter in USA / Europe are colder; because water absorbs more coldness / releases extra heat - that coldness is spread south by the currents: http://globalwarmingdenier.wordpress.com/midi-ice-age-can-be-avoided/

    ReplyDelete