Guest Post By Walter Dnes
|HadCRUT4 2016/06||+0.777 (based on incomplete data)|
The Data Sources
The latest data can be obtained from the following sources
- HadCRUT4 http://www.metoffice.gov.uk/hadobs/hadcrut4/data/current/time_series/HadCRUT.220.127.116.11.monthly_ns_avg.txt
- GISS http://data.giss.nasa.gov/gistemp/tabledata_v3/GLB.Ts+dSST.txt
- UAH http://vortex.nsstc.uah.edu/data/msu/v6.0beta/tlt/tltglhmam_6.0beta5.txt
- RSS ftp://ftp.ssmi.com/msu/monthly_time_series/rss_monthly_msu_amsu_channel_tlt_anomalies_land_and_ocean_v03_3.txt
- NCEI https://www.ncdc.noaa.gov/cag/time-series/global/globe/land_ocean/p12/12/1880-2016.csv
When this post was submitted, HadCRUT4 was only available through April 2016. The other 4 data sets were available through May 2016.
About the Data Sets
There’s no nice way of putting it. While my projections for 4 of the data sets (HadCRUT4/GISS/RSS/NCEI) were reasonably close, the UAH data set projections have busted badly. Looking into the situation, I noted that RSS data covers 82.5°N to 70°S and that UAH data covers 85°N to 85°S. I was using NCEP/NCAR data for 90°N to 90°S to correlate with all 5 data sets. That worked for HadCRUT4, GISS, and NCEI, and somehow for RSS. But not for UAH. Except for the latitude bands 90°N-88.75°N and 90°S-88.75°S, NCEP/NCAR data is done in 2.5 degree grids. I’ve modified my “global average” calculation program to generate 3 sets of anomalies…
- global 90°N to 90°S
- RSS 81.25°N to 68.75°S
- UAH 83.75°N to 83.75°S
This is as close a match as possible for the satellite sets. It may not be a perfect match, but it’s a much better “apples-to-apples” comparison than using 90°N to 90°S. How much better? We’ll find out in the next few days.
Here’s one qualitative prediction. I’ve been following various sources that show major cooling in Antarctica in June. Because it only covers down to 70°S, RSS will miss that cooling. Net result will be that RSS will show less cooling May-to-June than UAH. The surface data sets may miss it as well, due to their poor coverage of Antarctica, as indicated in the images below. My RSS-specific anomaly projects a minimal drop for June. But the surface data sets are assumed to cover the entire globe. This implies that the surface data sets (HadCRUT4/GISS/NCEI) may come in a bit warmer than my projections.
Global coverage (or lack thereof)
UAHv6, which covers 85°N to 85°S looks like the best data set as far as global coverage is concerned, followed by RSS with coverage 82.5°N to 70°S. The other surface and sea data sets are no better, and arguably worse. For example, from the GISS site:
Looks great? But select ocean data “none” and smoothing radius 250 km for May 2016 and click on “Make Map”. Try “Robinson” and “Polar Orthographic” projections. Land areas in grey are missing. Not only is GISS missing a lot of the polar regions, but also most of Africa and the Arabian Peninsula, plus chunks of South America and Australia.
NCEI is no better. May land data is available here.
When combined with ocean data, that somehow translates into this:
HadCRUT4 has similar coverage as shown on their website. They use 5 degree grid squares, which are squares 556 km on each side at the equator. The larger grid squares give the illusion of better coverage, but don’t affect the reality
There are holes in the data regardless of which data set is used. The only question is how much data is missing. It would also be interesting to see the data points and smoothing used for sea surface temperature anomalies.