the Air Vent

Because the world needs another opinion

Sat/Ground Correlation

Posted by Jeff Id on February 20, 2010

On a whim, I’ve been looking at sorting surface station data for quality according to UAH satellite data.  It’s a process which, if done right, could have either validity or be used as a method cherry picking – [sarc]an advantage unique to climatology[/sarc] .  In messing around I sorted ground data against the final UAH global average, throwing out stations below a correlation threshold.  As expected, I was able to generate beautiful UAH curves with every wiggle and shape by selecting the ‘right’ temperature stations. Only somewhat surprisingly the trends returned were double the UAH trend – below.

This was created by correlating to UAH and rejecting anything with corelations below a pre-set threshold.

Anyway, the results above are nothing but junk that happens to have the same shape as UAH.  I began to look at sorting according to a gridded UAH series. On NicL’s suggestion, I correlated each stations gridcell to the surrounding gridcells in order to account for weather patterns.  If the prevailing winds on a sea shore station predominantly carry inland air across the station, you might correlate better to a gridcell which has more inland area.  The next graph is the gridded GHCN global trend since 1978 – no sorting involved.  One thing I noticed is that the blue color stations predominantly sit along shorelines.   The second plot below is the same data sorted by removing anything with lower than 0.6 correlation.

GHCN Trends since 1978

Trends Since 1978. Correlations less than 0.6 removed.

Most of the light blue stations correlated poorly to UAH.  There is almost a band of stations across the equator +/- 20 degrees which lost all stations based on poor correlation to UAH.

Remember, in a recent post NicL also noted the poor correlation of UAH to sea surface temperatures.

Correlation to Sea Surface Temperature

This makes some sense since while SST affects the lower troposphere, the satellites measure a substantial air thickness and it would take time to conduct/convect the heat from the sea surface through the whole thickness.

I then plotted the correlation vs latitude.

Correlation vs Latitude GHCN UAH since 1978

Regulars here may remember John Christy’s amplification factor between UAH and surface measurements.  He recommends a value of 1.2 times ground for everything except the tropics +/- 20 degrees where the value is 1.3.  This represents the dampening factor of the decadal variance.  If you’re familiar with the effect, I now wonder if the tropical difference in amplification factor is a result of the difference in proximity of the tropcial stations to the ocean.  IOW, looking at the geography in equatorial regions, many of the stations are positioned closer to the ocean such that trade winds would cause additional dampening of surface level measurements in comparison to the lower troposphere sat measurements.   The increased tropical surface temperature dampening effect may be due to nothing other than geography.

7 Responses to “Sat/Ground Correlation”

  1. Al said

    Are you able to get an entire time series out for an individual gridcell out of the satellite data?

    That is, daily temperatures from 1978-now(ish). Something that can be compared to daily temperatures instead of to yearly temperatures.

  2. Kenneth Fritsch said

    I would mke 2 observations:

    1. The correlations, that you show, overall are poor.

    2. I have looked at many GHCN adjusted surface station temperature anomaly time series within a 5 x 5 degree grid and found that while the time series appear to conform in the features and even breakpoints, the trends can vary greatly. That is the time series can have the same ups and downs and peaks and valleys but with different amplitudes.

    I have seen this conformation of peaks and valleys used in dendroclimatology to make a point that the tree ring widths and densities of various and different trees are reacting similarly to the same climate events. The problem that I have seen with this view (besides not being able to pin down what kind of climate events are being reacted to although volcano eruptions are frequently used and as temperature indicators) is that, like here when comparing station time series, it is the differing amplitudes that make the differences in trends.

  3. Layman Lurker said

    What about tropical correlation with a lag factor built in?

  4. Roy Lofquist said

    Seems I heard someplace that satellite measurements are calibrated using surface measurements. Something loopy here?

  5. MikeN said

    Correlations to an increasing trend should filter out those with high slopes, so double the trend shouldn’t be too surprising, right?

    You should correlate with global UAH, because stations have a teleconnection.

  6. RomanM said

    Jeff, for what it’s worth, I have written up a post on a method for combining station records to calculate a grid cell time series. You can find it at .

    It appears to have some decent properties.

  7. Cement a friend said

    Jeff, I am not sure if you look at WUWT. There is a post from DR Roy Spencer on a very similar project. I made a comment which I will not repeat because of space. However, I made mention that Warwick Hughes has a post about CRU and GHCN station data. I have looked at the some of the Southern Hemisphere Australian data and I agree with his comment in the TR027 link
    “Readers should email CDIAC and ask for copies of TR022 and TR027. The USA DoE foisted this stuff on the world.

    However – for the SH I have a full station list with GHCN populations added by me. Gives you a fair idea how many city stations were used. Readers can judge for themselves the veracity of the Jones et al statement on p1216 of Jones et al 1986b, where they state that “… very few stations in our final data set come from large cities.” This glib and lulling statement is detached from the reality that 40% of their ~300 SH stations are cities with population over 50K.”

    Maybe if enough US citizens ask for copies of TR022 & TR027 a message will get through to the administration.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: