BEST

SCOPE

This page contains notes and validation information for Berkeley Earth Surface Temperatures (BEST). The latest 2013 version is assumed in what follows, unless stated otherwise.

BEST data and documentation is available from:  https://berkeleyearth.org/

BEST is designed to be the “right answer” for the past variations of regional-average monthly-average daily maximum, minimum and mean temperatures. There may be some doubt about the “right answer” part, especially before the 20th century, but in general there should be less doubt about the ability of BEST to provide a good “reference series” for many land locations on Earth.

I use the following MATLAB to read BEST text files, created from the “Data Table” information available via the link above, after header lines have been removed, the ts array giving the monthly temperature anomalies:

[ys, ms, ts, c4, c5, c6, c7, c8, c9, c10, c11, c12] = …
textread(strcat(Folder,file),’%u %u %f %f %f %f %f %f %f %f %f %f’);

Partial validation results are presented on this page for regions that have DIYMETANALYSIS regional average temperature variations, currently showing only 12-month moving averages. Results for all individual months will be added later.

LATEST MONTHLY AVERAGE TEMPERATURES: June 2019

AUSTRALIA

So far I have found BEST data to be excellent for Australian Tmax and Tmin variations after around 1910, the date when many stations began temperature measurements.

Alice Springs Tmax

The first example is a comparison between my Central Australia Tmax of blog post “EXAMPLE 09”, and BEST Alice Springs Tmax. In the following plot I have subtracted my regional average temperature deviations from the BEST data (in blue), and compared with my regional moving average (in red).

Albury/Rutherglen Tmin

The second example for Australia is minimum temperatures from the region near the towns of Albury and Rutherglen:

Perth Tmax

The third Australian example is Perth Tmax:

NEW ZEALAND

BEST for the country of New Zealand, Tmax data, is badly wrong before around 1942. This can be seen in the “wiggle test” shown in the following figure, showing monthly average Tmax temperatures for January and July. The BEST data in blue follows the year-to-year variations of raw monthly Tmax data from around 1942, but not before that date.

One factor that is likely to be implicated is the very small number of stations in the GHCNMv3 Tmax database, the source of most of the data used, some comes from GHCN daily. The list of NZ stations in GHCNMv3 with Tmax data is very short:

50793012000 -35.1300 173.2700 87.0 KAITAIA
50793292000 -38.6500 177.9800 5.0 GISBORNE AERO
50793309000 -39.0200 174.1800 32.0 NEW PLYMOUTH
50793615000 -42.7200 170.9800 40.0 HOKITIKA AERO
50793780000 -43.4800 172.5200 37.0 CHRISTCHURCH
50793844000 -46.7000 168.5500 4.0 INVERCARGILL

It is possible that the BEST methodology uses more distant stations when necessary, many of which are small islands with almost purely maritime weather.

More to follow …