Under the Volcano, Over the Volcano

Guest Post by Willis Eschenbach

In 2006, I lived for a year in Waimea, on the Big Island of Hawaii. From my house I could see the Mauna Loa Observatory (MLO). This observatory is the home of the longest continuous series of CO2 measurements we have. The recording station was set up by Dave Keeling in 1959, and has operated continuously ever since.

Figure 1. Mauna Loa Observatory ( 19.536337°N, 155.576248°W)

Here’s a view of the observatory:

Every time the subject of CO2 measurements comes up, people raise all kinds of objections to the Mauna Loa measurements. So I thought I’d start a thread where we can discuss those objections, and perhaps dispose of some of them.

Here are the objections that I hear the most:

1. The Mauna Loa results don’t measure the background CO2 levels.

2. You can’t get accurate CO2 measurements from samples taken on the side of an active volcano that is outgassing CO2.

3. The measurements from Mauna Loa are not representative of the rest of the world.

4. What about the Beck data, doesn’t it contradict the MLO data?

5. Keeling chose a bad location.

Before we get into those issues, let’s start by looking at the local meteorological conditions at the site. Mauna Loa is at an elevation of 3397 metres (11,140 ft) on the side of a 4,170 metre (13,680 ft) volcano way out in the middle of the Pacific Ocean. Because it is on an island, it gets the “sea breeze” in the daytime, and the “land breeze” in the nighttime.

These winds are caused by the differential heating of the land and the sea. Land heats up much faster than the ocean. So during the day, the warmer land heats the air, which rises. This rising air is replaced by air moving in from the surrounding ocean, creating the “sea breeze”.

At night, the situation is reversed. The land is cooler than the ocean. This cools the air. The cool air runs downhill along the slopes of the island and out to sea, creating the “land breeze”. Here’s a drawing of the situation:

Figure 2. Day and night breezes at Mauna Loa.

Now that we understand what is happening at Mauna Loa, let’s look at the objections.

1. The Mauna Loa results don’t measure the background CO2 levels. As you might imagine from Fig. 2, the CO2 measurements are taken only at night. Thus, they are measuring descending air that is coming from thousands of feet aloft. This air has traveled across half of the Pacific Ocean, so it is far from any man-made CO2 sources. And as a result, it is very representative of the global background CO2 levels. That’s why Keeling chose the site.

2. You can’t get accurate CO2 measurements from samples taken on the side of an active volcano that is outgassing CO2. This seems like an insuperable objection. I mean, Mauna Loa is in fact an active volcano that is outgassing CO2. How do they avoid that?

The answer lies in the fact that the volcanic gasses are very rich in CO2. At night, they are trapped in a thin layer near the ground by a temperature inversion.

To detect the difference between volcanic and background CO2, the measurements are taken simultaneously from tall towers and from near the ground, at intervals throughout the night. Background CO2 levels will be around 380 ppmv (these days), will be steady, and will be identical at the top and bottom of the towers. Volcanic gasses, on the other hand, will be well above 380 ppmv, will be variable, and will be greater near the ground than at the top of the towers.

This allows the scientists to distinguish reliably between volcanic and background CO2 levels. Here is a description of the process:

Air samples at Mauna Loa are collected continuously from air intakes at the top of four 7-m towers and one 27-m tower. Four air samples are collected each hour for the purpose of determining the CO2 concentration. Determinations of CO2 are made by using a Siemens Ultramat 3 nondispersive infrared gas analyzer with a water vapor freeze trap. This analyzer registers the concentration of CO2 in a stream of air flowing at ~0.5 L/min. Every 30 minutes, the flow is replaced by a stream of calibrating gas or “working reference gas”. In December 1983, CO2-in-N2 calibration gases were replaced with the currently used CO2-in-air calibration gases. These calibration gases and other reference gases are compared periodically to determine the instrument sensitivity and to check for possible contamination in the air-handling system. These reference gases are themselves calibrated against specific standard gases whose CO2 concentrations are determined manometrically. Greater details about the sampling methods at Mauna Loa are given in Keeling et al. (1982) and Keeling et al. (2002).

Hourly averages of atmospheric CO2 concentration, wind speed, and wind direction are plotted as a basis for selecting data for further processing. Data are selected for periods of steady hourly data to within ~0.5 parts per million by volume (ppmv); at least six consecutive hours of steady data are required to form a daily average. Greater details about the data selection criteria used at Mauna Loa are given in Bacastow et al. (1985). Data are in terms of the Scripps “03A” calibration scale.

There is a more detailed description of the measurement and selection process here.

As a result, the Mauna Loa record does accurately measure the background CO2 levels, despite the fact that it is on an active volcano. The samples that are identified as volcanic CO2 are not thrown away, however. They are used for analyses of the volcanic emission rates, such as this one (pdf).

3. The measurements from Mauna Loa are not representative of the rest of the world. Well, yes and no. The concentration of atmospheric CO2 varies by month, and also by latitude. Here is a “carpet diagram” of the changes by time and latitude.

Figure 3. A “carpet diagram” of CO2 distributions, by time and latitude.

Note that the swings are much greater in the Northern Hemisphere. Presumably, this is from the plants in the much larger land area of the Northern Hemisphere. However, the difference between the annual average of the Northern and Southern Hemispheres is small. In addition, there are smaller daily variations around the planet. An animation of these is visible here, with day by day variations available here.

Figure 4 shows is a typical day’s variations, picked at random:

Figure 4. Snapshot of the variations in tropospheric CO2. Note that the range is small, about ±1% of the average value.

In general, the different global records match quite closely. In addition to the Mauna Loa observatory, NOAA maintains CO2 measuring stations at Barrow, Alaska; American Samoa; and the South Pole. Here is a comparison of the four records (along with two methane records):

Figure 5. Comparison of the CO2 records from the four NOAA measuring sites.

As you can see, there is very little difference between the CO2 measurements at the four stations – two in the Northern Hemisphere, two in the Southern, two tropical, and two polar.

4. What about the Beck data, doesn’t it contradict the MLO data? In 2007, Ernst-Georg Beck published a paper called “180 Years Of Atmospheric CO2 Gas Analysis by Chemical Methods” (pdf).  In it, he showed a variety of results from earlier analyses of the atmospheric CO2. In general, these were larger than either the ice core or the MLO data. So why don’t I believe them?

I do believe them … with a caveat. I think that the Beck data is accurate, but that it is not measuring the background CO2. CO2 measurements need to be done very carefully, in selected locations, to avoid contamination from a host of natural CO2 sources. These sources include industry, automobiles, fires, soil, plants, the list is long. To illustrate the problems, I have graphed the Beck data from his Figure 13, against the Law Dome ice core data and the MLO data.

Figure 6. CO2 data from a variety of sources. White crosses are MLO data. Three separate ice core records are shown. Photo is of Mauna Loa dusted with snow (yes, it snows in Hawaii.) PHOTO SOURCE

There are several things to note about this graph. First, there is good agreement between the Law Dome ice core data and the MLO data over the ~ two decade overlap. Second, there is good agreement between the three separate Law Dome ice core datasets. Third, both the ice cores and the MLO data do not vary much from year to year.

Now look at the various datasets cited by Beck. Many of them vary quite widely from one year to the next. The different datasets show very different values for either the same year or for nearby years. And they differ greatly from both the ice core and the MLO data.

Because of this, I conclude that the Beck data, while valuable for showing ground level CO2 variations at individual locations, do not reflect the background CO2 level of the planet. As such, they cannot be compared to the MLO data, to the ice core data, or to each other.

5. Keeling chose a bad location. I would say that Keeling picked a very good location. It not only allows us to measure the background CO2 in a very accurate manner, it provides invaluable information about the amount of CO2 coming from the volcano.

My conclusion? Most of the records in the field of climate science are short, spotty, and not very accurate. We have little global historical information on ocean temperatures, on land temperatures, on relative humidity, on atmospheric temperatures, on hurricane occurrence and strength, or on a host of other variables. By contrast, the Mauna Loa CO2 records are complete since 1959, are very accurate, and are verified by measurements in several other locations.

I’m about as skeptical as anyone I know. But I think that the Mauna Loa CO2 measurements are arguably the best dataset in the field of climate science. I wouldn’t waste time fighting to disprove them, there are lots of other datasets that deserve closer scrutiny.

[UPDATE] A reader below has added another question, viz:

6. What about Jaworoski’s claim that the ice core data has had its age “adjusted”?

Jaworoski argues that the age of the air in the ice cores has been “adjusted” to make it align with the modern data. He says, for example, that the Siple ice core data has been moved forwards exactly 83 years to make them match the Mauna Loa data.

Dating the ice core data is problematic. We can date the ice itself pretty accurately, through counting layers (like tree rings) and through studying various substances such as volcanic dust that is trapped in the ice. However, dating the air is harder.

The difficulty is that the air is not trapped in the ice immediately. The pores in the “firn”, the snow that falls annually on top of the ice are open. Air can flow in and out.

As more and more snow falls over the years, at some point the pores close off and the air is trapped. So how long does it take for the pores in the firn to seal off?

Unfortunately, as in so many areas of climate science, the answer is … “depends”. It depends inter alia on how much snow falls every year, how much of that snow sublimates (changes from a solid to a gas) every year, and even the shape and size of the individual snowflakes.

The end result of all of this is that we end up with two ages for any given thin slice of an ice core. These are the “ice age” (how old the ice itself is), and the “air age” (how old the air trapped in the ice is). The ice is always older than the air.

The main variable in that is thought to be the annual snowfall. Unfortunately, while we know the current rate of annual snowfall, we don’t know the historical rate, particularly tens of thousands of years ago. So we use the concentration of an isotope of oxygen called “d18O” to estimate the historical snowfall rate, and thence the firn closing rate, and from that the air age.

Sounds a bit sketchy? Well … it is, particularly as we go way back. However, for recent data, it is much more accurate.

So to bring this back to real data, in the ice core data I showed in Fig. 5, the air is calculated to be 30 years younger than the ice for cores DEO8 and DEO8-2, and 58 years younger for the DSS core. Is this correct? I don’t know, but I do know that there are sound scientific reasons for the “adjustment” that Jaworowski objects to .

Finally, the existence of a thirty to sixty year difference in air and ice age doesn’t make much difference in the pre-industrial levels of CO2. This is because prior to about 1800, the level is basically flat, so an error in the air age dating doesn’t change the CO2 values in any significant manner.

FURTHER INFORMATION:

MLO Home Page

MLO Data Links

Greenhouse Gas & Carbon Cycle Research Programs

Trends in Atmospheric Carbon Dioxide

MLO Webcams

How we measure background CO2 levels on Mauna Loa