Annual Temperature Plots - Histograms

Annual Temperature Plots
Histograms - Oceans

My GHCN Temperature Plotter tool provides a number of tools to help analyze the provided data. The basics on how to use the Histogram tab are covered on another page - this page concentrates on the ocean data and explains why it is fairly useless.

There are many reasons that I consider the provided ocean data to be worthless. For instance, before the satellite era

The methods of determining temperature changed over time - buckets, engine intakes, etc.
No specific locations were used - just wherever a ship happened to be
There was no regular time of sampling - a few times a month, any time of the day
And lots more

As a result, the error bars are enormous, yet, in the available data, there is no indication of the size of the uncertainty.

This page ignores those and concentrates on just the data itself.

Basic Histograms | Compare with land | Compare past to present | Anomalies | Comments

Basic Histograms

To reproduce these histograms

Select only the ocean
- Select GHCN adjusted+ocean
- Select the By country radio button
- Click on (or near) any ocean (blue) site (dot)
On the Map Trends tab
- Set the date range (trend period) to 1960 to 2014
On the Histograms tab
- Set the Baseline bin to 0.01 (default = 1.0)
- Set the Latitude and Longitude bins to 10 (defaults = 2 and 3)
- Click each of the radio buttons to produce these plots

Ocean Histograms


South weighted	East Pacific - Atlantic - Indian - West Pacific

Since only anomaly temperatures are provided, I expected them to all be centered ON zero, not around zero!

To map the various baseline temperatures,

On the Map Data tab
- Select Baseline
- Set the Delta value to 0.02°C/bin
On the Bin Colors tab
- Select either 5 levels or 7 levels

Ocean Baseline Temperature color dot map 7 Color Levels - 0.02°C/bin

Compare with land

We are often told that "the heat is hiding in the oceans". That implies (to me) that the oceans are warming slower than what the models predict. That's hard to prove (or disprove), but we can compare the rate of heating the ocean to the land masses.

Ocean (blue) vs Land (red) Histograms 1900-2014
	This example uses A Trend bin of 0.01 All the default filters to get only the long term data The adjusted land data (388 stations) 333 ocean stations

On a daily basis, the land warms (and cools) much faster, and much more, than the ocean. In fact, this is the major driver of coastal weather - off-shore breezes during the day, on-shore breezes at night. However, over a multiyear time period, I don't know of any theory that suggests that the land will warm faster than the ocean.

As a result, the fact that the two curves are so different indicates (to me) that there is a problem with the data.

It is also interesting to compare the number of years of data per site - I don't know what this indicates, I just thought it was interesting.

Number of years of data Ocean (blue) vs Land (red)
	This example uses NumofYrs bin = 5 Only the default baseline filter: 1961-1990 4,561 adjusted land data stations 884 ocean stations

Compare past to present

Perhaps looking at the entire dataset is the wrong approach. We often hear "Its heating faster now than in the past".

This one is a bit tricky because the number of sites changes over time. As a result, I have selected only those sites that have a long record.

Ocean temperature trend 1900-1920 (blue) vs 1990-2010 (red)
	This example uses A Trend bin of 0.05 All the default filters to get only the long term data 333 ocean stations

Well, that obviously supports the claim that the change in temperature is greater now.

When we also include the land data, that strongly supports the "warming faster now" claim.

Ocean (blue) and Land (red) temperature trends
1900-1920	1990-2010

Unlike the histogram for the entire dataset (1900-2014), when 20-year blocks of time are viewed, the ocean and land data appear to have about the same warming rate. Again, (to me) this indicates a data problem - the long term and short term plots should agree.

Anomalies

For comparison, this plot shows the temperature anomalies for the ocean and adjusted land stations using only the baseline filter. The number of sites per year is shown in the lower graph.

Ocean (blue) and Land (red) Anomalies Trend lines and slopes for 1980-2014

Notice that the plots have almost identical values until about 1980 - after that, the land warms at about twice the rate of the ocean. Again, I don't know any theory to support this except for bad data.

When plotting histograms with all the data for shorter time frames, some of the bad data becomes obvious - for instance, at least one land site had a bogus cooling trend of more than -20°C/decade! The following histograms are plotted with fixed scaling to hide that nonsense and to show the useful details. (Note that these plots use different date ranges than those in previous sections because the anomaly plot suggests that when all the data is used, these ranges might be better.)

Ocean (blue) and Land (red) temperature trends - Bin size = 0.05°C/decade
1920-1940			1980-2014

	Included	Excluded		Included	Excluded
Ocean	860	24	Ocean	884	0
Land	2075	2486	Land	4264	297

This color dot map of the ocean and land data shows the temperature trend for each site from 1980 to 2014 (where the slopes differ in the previous anomaly trend image). Since I strongly dislike analyses which show only warming or cooling, I have used green to indicate sites with almost no temperature trend. If you prefer a different value, the guardband of ±0.05°C/decade can be modified in the application.

Green indicates sites with a trend of ±0.05°C/decade
Red indicates sites with a warming trend of greater than 0.05°C/decade
Blue indicates sites with a cooling trend of more than 0.05°C/decade
White indicates insufficient data (fewer than 3 samples in the date range)

Temperature trend color dot map - Adjusted Temperatures 1980-2014

So much for "Global" warming - while most of the dots (sites) are red (warming), there are plenty of green (no change) and blue (cooling) sites.

In South-East Australia, there are a lot of sites with strong cooling (blue) right next to sites with strong warming (red). A small sample shows that those with a strong cooling have no data after 1990! Why so many sites stopped reporting data at that time is a mystery (to me anyway). At any rate, Berkeley Earth appears to have data for those years and GHCN does not (based on a small sample). When I filtered the data to require 8 years of data between 2000 and 2010, the number of Australian sites went from 347 to 57. For the entire planet, the number of adjusted land sites went from 4561 to 1622 and most of the blue (cooling) sites went away (which is why I make filters like that available).

Comments

Is it heating faster now than in the past?
Are the very obvious trends simply noise?
Is the data worthless?

Unfortunately, I don't have answers to these questions. Since the provided data seems to support whatever you want, I tend to support the position that no one knows what is really happening.

At any rate, it is my opinion that the ocean data is more the result of a model than anything based on measurements and, as a result, should not be used for anything beyond entertainment.

Author: Robert Clemenzi