Thursday, July 15, 2010

Revisiting Bolivia

Back in April I looked at the "Bolivia effect". This was attributed to the absence of GHCN data from Bolivia since 1990. This does create a gap in the temperature data; there are stations just over the border.

The conclusion was that there was no real evidence of a major problem, but perhaps some uncertainty remained. Now with GSOD data, which is available for Bolivia through to present, we can look again.



I'll do a side-by side GSOD-GHCN presentation. First is the anomaly plot and trend for a region within 1200 km of La Paz, since 1990. This can be calculated for both sets, even though GHCN has no data within the country.



GSOD Within 1200 km of La Paz


GHCN Within 1200 km of La Paz


GSOD


GHCN


GSOD


GHCN


Both plot and trend are similar. GHCN does a good job with the region, even lacking data within the country. Next is a plot of GSOD data from Bolivia data. There are no corresponding GHCN results, but it is quite similar to the GHCN result for the 1200 km circle.

4 comments:

  1. Noise, meet noise that's correlated to you.

    There have been some CLIMAT reports for Bolivian stations in the last couple years, so these will probably get picked up by the GHCN sooner or later.

    One interesting and more focused thing to do would be to see how stations that are near each other, but are in very different physical environments (mountain vs valley vs coastal) correlate with each other. I think this is the basis of some of the complaints. Of course, we didn't need GSOD or Bolivia to do this; probably the US has plenty of examples.

    ReplyDelete
  2. Also, I think some of these comparisons would be better presented on a single plot, instead of side by side. Maybe that's more work for you, but it's less work for us. As you wish.

    ReplyDelete
  3. CE, I have to admit that ease of production is a factor in this choice. But I think also that the predictability of the graph format is a plus for readers - maybe I'm rationalising. More strain on the eyes but less on the brain?

    Not much I can do about the moise. It's true that you'd expect the larger station numbers with GSOD should give less noisy signal, but it doesn't seem to. Maybe that reflects data quality.

    ReplyDelete
  4. T'isnt actually noise, Nick. If add more and more stations in the region, and you still get the same fluctuations when you average them together, that's the actual variability at that place. Real weather is climate noise, but it's still physically real.

    That may go without saying, but just to be clear. But that's why I wanted an overlay, to see how well the 'noise' matches up.

    ReplyDelete