Joint post with Erik Swenson who studied computational physics and is based in California.
OCO-2 is NASA’s latest and most advanced instrument for measuring the distribution of CO2 from space. Launched in July 2014, NASA published a map in January of this year summarising data from 1 October to 11 November 2014 that I covered in a post titled CO2 – The View From Space. But then there has been several months of silence.
The silence was broken recently when NASA dumped a year of OCO2 data onto one of their servers. They have not provided fresh maps, just raw data in a format not accessible to the layman. Erik Swenson writing at WUWT has accessed and processed the data and produced a series of maps which tell an interesting story.
A good starting point is to look at one of Erik’s maps for a 52 week average of data and compare this with global forest cover. The 52 week average is understood to provide a picture of net sources and sinks of CO2 (upper map). Many of the net sources appear to correlate with global forest cover (lower map) which is counter to intuition that global forests have acted as a net sink for anthropogenic CO2.
Forest map from the Japanese Space Agency
I want to try and keep this post short and simple so that many can follow the story, but that in itself is a challenge. The first task will be a brief examination of quality control. How do we know Erik’s maps, based on the NASA data, are reliable? The second task will be a few comments on the carbon cycle and the HUGE upheavals in CO2 around the globe across the annual cycle, that, when averaged give rise to the ordered net sources and sinks map shown at the top of this post. The third task will be to make observations about the net sources and sinks and to speculate about what these observations might mean.
Task 1: Data Quality
There are two important aspects of data quality. The first is what NASA measures from space and how reliable this represents CO2 in the atmosphere. The second is how reliably Erik has managed to capture this data and project it into the maps he has produced. The data are acquired by a polar orbiting satellite and because of various physical limitations, the data N and S of 60˚ are less reliable than the data between these latitudes. NASA has released two data sets. The OCO2 LITE data, that Erik has used, is regarded as the most reliable.
In his WUWT article, Erik was very obliging in responding to requests. And one request I made was that he produce a map for 1 October to November 11 2014 on comparable scale to that previously published by NASA. Below are Erik’s (upper) and NASA’s (lower) maps:
There is a high degree of correspondence between the two but also some discrepancies, most notably in the N Atlantic and N Pacific. The areas of agreement are sufficiently good to warrant Erik’s methodology is good. The areas of discrepancy are most likely explained by different versions of input data being used.
Task 2: CO2 Convulsions Across the Annual Carbon Cycle.
The maps below each summarise 6 weeks of data, the minimum time required for the satellite to gather representative global cover. This also follows the NASA convention. Erik produced 8 such maps for 8 time slices which can be viewed at WUWT and at the end of this post. Note that the scale bars have been changed on the maps below relative to those published on WUWT (an error was found that has since been corrected). The picture is very complex and in interest of trying to simplify this I show only 4 maps below. My instinct was to show the maps that spanned the winter solstice, spring equinox etc. But it is in fact the maps in between these seasonal markers that show the more extreme behaviour that I have shown below. NH = northern hemisphere.
1 October to 11 November = NH Autumn
1 January to 15 February = NH Winter
1 April to 15 May = NH Spring
1 July to 15 August = NH Summer
If you are struggling to make sense of the maps, don’t worry, I think everyone will be. But as a starting point, it is important to appreciate the enormous natural CO2 fluxes involved.
The IPCC and climate science community accept that annual manmade emissions are tiny compared with the natural fluxes and stores. It is unlikely that a single year of OCO2 data would show any emissions CO2 against such a large natural flux.
Task 3: Observations and Interpretation
A good starting point for the discussion is to compare the 4 time slices with the Mauna Loa CO2 data. The annual cycle at Mauna Loa has long been linked to photosynthesis, a hypothesis that has never made complete sense to me. The mosaic below rationalises the OCO2 maps with atmospheric observations. But it is less easy to link the observations to biological and natural processes.
It is convenient to begin the discussion with the 1 April to 15 May NH Spring time slice (lower left quadrant labelled May). This shows the NH “awash with CO2”. The SH in Autumn is largely neutral. The NH highest concentrations are over the rain forests of the Amazon and Congo, SE Asia and above the Boreal forests of Canada and Russia. Alarmists may want to imagine elevated CO2 over E United States, Europe, India and China. But manmade emissions are tiny compared with the natural flux and should be in decline at this time of year. I would not expect manmade emissions to be visible.
In this time slice the NH growing season is just getting under way and one may have expected plant growth to be drawing down CO2 while the opposite is clearly happening. I will hypothesise that we are observing soil processes and decay of organic matter. This could be linked to trees pumping organic acids into soils or microbial activity. It is more than a little curious that the tropical forests N of the Equator also pulse CO2 at this time since they lack a seasonal driver. Taking a step back to the 1 January to 15 February time slice (top right) we see that the pulse of NH forest emissions begins in mid-winter.
I am happy to entertain other ideas from informed commenters. Elevated CO2 over the NH oceans I will assume is linked to circulation from processes taking place over the forests.
Moving on to the next time slice, 1 July to 15 August (NH Summer, bottom right) we see a very dramatic change where the high latitude Boreal forests turn from source into sink pumping down atmospheric CO2. This can be reconciled by photosynthesis and tree growth. Notably the tropical forests of the NH are largely neutral at this time. This time slice represents the depths of the SH hemisphere winter and while much of the land is neutral, the southern Ocean has turned into a net CO2 source, counter to the intuition that cold water should absorb more CO2.
The 1 October to 11 November time slice (NH Autumn, SH Spring, top left labelled Nov) shows a time of CO2 sequestration across the globe, consistent with the low point of the Mauna Loa cycle. CO2 is evidently being pumped down by high and low latitude oceans. But a surprising observation is reduced CO2 over or close to many of the Earth’s deserts. I am unable to offer an explanation.
To conclude I return to the 52 week summary map posted at the top of this post. Below I have added numbers to the main features that are described below.
- Much of the land and ocean area is coloured green reflecting that it is neither a net source or sink. Considering the annual carbon cycle convulsions described above, this is somewhat surprising but also gives a sense of security in data quality.
- The SH on average appears to be neutral with some positive and some negative residuals.
- Large segments of the NH are shaded neutral green but the forested land masses appear to be a major net source of CO2
- The global thermohaline circulation emerges in the N Pacific to the S of Alaska and this positive residual may be linked to the emergence of deep, carbon rich, low pH ocean water.
- The tropical forests of the Amazon and west central Africa are a net source of CO2.
- The tropical forests of SE Asia, China, Korea and Japan are a significant net source of CO2.
- The Boreal forests of Canada and the eastern seaboard forests of the USA are a significant net source of CO2.
- Elevated residuals over the N Atlantic and N Pacific have the appearance of offshore plumes sourced from the onshore forests.
- The southern continental extremities of S America, Africa and Australia appear as minor net sinks for CO2.
- The high latitude southern ocean appears to be a net sink for CO2.
I realise that many readers will be looking at positive residuals over the eastern USA, Europe, China, S Korea and Japan and be thinking that this must be proof positive of manmade emissions. Here, socio-economic history plays a cruel trick. Before we started using coal, industrial society was rooted in the areas where there was ample timber. The forest cover map up top shows that the eastern USA, Europe, China, S Korea and Japan are all heavily forested. The correlation here is with forest cover and not with Man. The positive residuals in the Amazon, Congo and over the Boreal forests show this to be the case. And the absence of residuals over much of the USA, S Africa, The Middle East, SE Australia and India shows that industrialised areas away from forests do not have positive residuals. The populated and industrialised areas of coastal Brazil have a neutral residual while the deep Amazon Forest appears as a net source.
To hammer home the point about the link between positive residuals and forest cover, the map below shows the two overlaid. I got lucky with the map projections which are exactly the same.
It goes beyond the scope of this post to try and explain why global forests appear to be a net source of CO2. This implies the forests are out of equilibrium. This may be linked to natural climate cycles and it would not surprise me in the least if forests go through cycles of sequestering and then emitting CO2. This does not necessarily mean that the forest mass is decreasing since it is possible that the CO2 is linked to soils. There is a large amount of research required to understand exactly what is going on and the addition of further data may assist this understanding. But OCO-2 is only scheduled to acquire data for two years.
This post is prepared in good faith based on the assumptions that the input data and Erik Swenson’s maps are accurate and reflect the dynamics of CO2 sources and sinks. If one of these three assumptions fails then we reserve the right to withdraw all of this commentary.
All the available maps are given below for reference purposes.
1 Oct to 11 Nov 2014
16 Nov to 31 Dec 2014
1 Jan to 15 Feb 2015
16 Feb to 31 Mar 2015
1 April to 15 May
16 May to 30 Jun 2015
1 July to 15 aug 2015
16 Aug to 22 Sep 2015