Posted by drj | Filed under Uncategorized
Along with other changes I’ve been making to ccc-gistemp I recently added a feature to remove GHCN stations that are in the contiguous US. I did that partly because GISTEMP’s Fortran has a similar feature, and partly because it allows me to do the following whimsical experiment:
Running ccc-gistemp with no USHCN data and dropping contiguous US stations from GHCN:
tool/run.py -p 'retain_contiguous_US=False;data_sources=ghcn,scar,hohenpeissenberg'
The -p allows a parameter from parameters.py to be set, without editing that file. retain_contiguous_US and data_sources are recently added parameters.
Onto the graphs!
Hmm, not much difference.
What about removing the ocean? Removing it should bring the differences due to land stations into sharper focus:
Tiny differences, not easily distinguishable from rounding.
What about zooming into the Northern Hemisphere, where the US is located?
Aha, some slightly less tiny differences appear which are not due to rounding. US citizens may be interested to note that their contribution to the recent trend is to lower it by a tiny amount (the trend is lower when US stations are included).
ccc-gistemp produces zonal anomaly files (because GISTEMP does). I’ve not really looked at them much before, but we can zoom in further on Zone 24N to 44N where most of the contiguous US is found. Note I had to rescale the next chart to fit the higher peaks in:
What above the next (GISTEMP) zone north, which includes the remaining part of the contiguous US but which has a slightly less dense station network and more volatile temperature swings:
The differences are all lost in the noise?
The US is only about 1½ percent of the Earth’s surface after all.